Karan's Notes

Ai's Mathmetica

Fri, 08 May 2026 07:19:51 GMT

https://davidbessis.substack.com/p/the-fall-of-the-theorem-economy?s=r&curius=1819

I’ll inspect the linked essay and summarize its central thesis, key claims, and takeaways, noting any access limits such as a paywall.

Thought for 1m 37s

I read it. Here’s the compact map.

Core thesis

David Bessis argues that AI may soon collapse the “theorem economy”: the academic system that rewards mathematicians mainly for proving new theorems. But that would not mean AI has “solved mathematics,” because the real product of mathematics is understanding, concepts, intuition, canonized structures, and clearer ways of thinking, not isolated theorem-statements or machine-checkable proof blobs. The essay is titled “The fall of the theorem economy”, subtitled “How AI could destroy mathematics and barely touch it,” and was published on April 21, 2026.

The argument in one sentence

AI could become extremely good at producing proofs before it becomes good at producing the thing mathematicians actually value most: humanly meaningful mathematical understanding.

Main moves in the essay

Bessis starts with a personal example: some of his most valuable mathematical contributions were not the formal theorem-statements themselves, but the definitions and conceptual frameworks that made the results almost inevitable. His point is that theorem-proving often looks like the hard part from the outside, while the deeper creative act is often finding the right language, definitions, and viewpoint.

He contrasts “official math” with “secret math.” Official math is the public image of mathematics as formal deduction from axioms to theorems. Secret math is the messy human practice of intuition, sense-making, concept-building, and learning how to see. He says the profession has historically rewarded the official version because theorems are easier to evaluate, even though the hidden version is often where the real value lies.

The essay’s anxiety comes from AI: if labs and the public treat math as a closed game like chess or Go, then an AI system that solves benchmark problems or produces formal proofs may be presented as having “beaten” mathematics. Bessis thinks that framing is dangerously wrong, because mathematics is not a finite adversarial game with a scoreboard; it is an open-ended human project of building understanding.

His three big caveats about AI-for-math

1. “Oceans.” Bessis says benchmark problems such as the First Proof questions may be “research-level,” but they are closer to technical lemmas than to deep mathematical breakthroughs. There are, in his phrasing, oceans between solving such problems, writing a serious paper, making a breakthrough, and producing a Fields-medal-level contribution.

2. Accretiveness. A proof can be correct but still fail to add much to mathematics if it is unintelligible and not reusable. This is where he discusses formalization, Lean, Mathlib, and the controversy over giant autoformalized proofs. His worry is not merely “can we verify it?” but “does it become part of the living mathematical library in a form others can understand and build on?”

3. The Overhang. He introduces “the Overhang” as the latent value inside the existing mathematical corpus: unnoticed connections, unused concepts, and partially buried ideas that could solve problems once connected. LLMs may be unusually good at harvesting this overhang because they can absorb far more of the literature than any individual mathematician. That could produce real discoveries, but also a grim dynamic where humans curate the corpus and AI systems capture the visible prize.

What he thinks mathematicians should do

Bessis thinks mathematicians need to “come clean” about what mathematics is really for. They should stop pretending, even implicitly, that the discipline is just a theorem factory. He wants the community to explicitly defend concept-building, canonization, problem-setting, reframing, simplification, intuition, and mathematical understanding as central outputs of the field.

He also suggests something like a Mathematical Intelligence Scale or Mathematical Automation Scale, analogous to levels of self-driving automation. The idea is to prevent misleading headlines like “AI solved math” by distinguishing between narrow theorem-solving ability and broader mathematical intelligence, including intelligibility and theory-building.

My read

The essay’s strongest point is that “proof” is not the same as “mathematics.” A verified proof can settle a proposition, but mathematics also depends on compression, explanation, reusable abstractions, aesthetic judgment, pedagogy, and shared mental models. Bessis is not denying that AI will be powerful; he is arguing that the public narrative may mistake a dramatic advance on one proxy for the conquest of the whole discipline.

The most interesting concept is the Overhang. It gives a plausible reason why AI might make impressive mathematical progress without yet having human-like mathematical taste: much “new” mathematics may consist of connecting existing pieces that no human has jointly internalized. That would still be valuable, but it would not automatically imply that AI has mastered concept creation or mathematical meaning.

The essay is pessimistic about the profession but not about mathematics itself. Bessis expects research mathematics to be disrupted, perhaps severely, but he thinks human mathematics survives because humans will still need understanding. His closing picture is not “AI kills math,” but “AI forces math to admit what it was really about all along.”

so what should one do??

Thought for 58s

One should not quit mathematics. One should quit trying to be a theorem factory.

The practical move is:

Optimize for understanding, not just proof production

If AI makes proofs cheaper, then the valuable human work shifts toward:

choosing the right problems, making the right definitions, explaining why something matters, building conceptual maps, connecting fields, and turning proof into understanding.

So the question becomes less:

“Can I prove something new?”

and more:

“Can I see what is going on better than others, and help others see it too?”

That is the part most worth cultivating.

For a student

Keep learning hard mathematics. Do not outsource the struggle too early. The struggle is how taste and intuition are formed.

But also start using AI as a mathematical sparring partner. Ask it for examples, counterexamples, alternate proofs, analogies, failed proof attempts, simpler cases, historical context, and “what is the hidden structure here?” Then verify everything yourself.

The mistake would be using AI to avoid thinking. The smart use is using AI to create more thinking surface area.

A good routine would be:

Study a theorem normally.
Before reading the proof, try examples.
Ask AI for intuition and possible proof strategies.
Compare with the textbook proof.
Write your own explanation in plain language.
Ask: “What concept made this theorem inevitable?”

That final question is the key.

For a researcher

Start treating “proof of theorem X” as only one kind of output.

Other valuable outputs include:

a better definition, a clean conjectural picture, a taxonomy of examples, a bridge between two literatures, a simplified proof, a formalized library, a survey that makes a field usable, or a computational playground that reveals structure.

Those things may become more important, not less.

Also: learn to work with proof assistants and AI tools. Not because they replace mathematicians, but because they may become the new lab equipment. A mathematician who can combine taste, formalization, computation, and exposition will be in a much stronger position than someone who only competes at hand-proof speed.

For someone choosing a career

The safest bet is not “pure theorem-prover.” The safer bet is:

deep mathematical maturity + computational fluency + communication + taste.

That combination travels well: AI research, cryptography, quantitative science, theoretical CS, economics, physics, biology, data science, formal verification, education, scientific software, and academic math itself.

Pure math is not dead. But a narrow career based on producing small publishable theorems may become more fragile.

What to cultivate now

Cultivate these five skills:

1. Taste
Which problems matter? Which definitions are natural? Which analogy is real and which is superficial?

2. Concept-building
Can you invent or identify the abstraction that makes many facts obvious?

3. Exposition
Can you explain a hard idea so clearly that others can use it?

4. Verification discipline
Can you tell when an argument, human or AI-generated, is actually correct?

5. Tool fluency
Can you use AI, computation, formal proof systems, and databases as extensions of your mathematical mind?

The dangerous traps

One trap is despair: “AI will do all math, so why bother?” That is too passive.

Another trap is denial: “Real math is sacred and AI will never matter.” That is probably wrong too.

The better attitude is:

AI may commoditize some proof work. So become excellent at the parts of mathematics that give proofs meaning.

The simplest answer

Do mathematics in a way that makes you less replaceable by a theorem generator.

That means: build taste, understand deeply, write beautifully, connect ideas, learn the tools, and focus on questions where the hard part is not merely proving a statement but knowing what the statement should be.

Mathematica + Lean + arXiv/MathSciNet-style literature graph + Jupyter/Wolfram notebooks + AI agent + simulator + tutor + verifier. this kind of platform doesn’t exist??

I’ll separate a complete integrated platform from today’s stitched-together ecosystem: many pieces exist, but the unified AI-native math workbench is still largely missing.

Thought for 2m 56s

Not really — not as one polished, public, end-to-end platform.

What exists today is a pile of powerful components, not the unified “mathematical research OS” you described.

Layer you named	Exists?	But…
Mathematica / Wolfram computation	Yes	Great computation + notebooks + AI assistant, but not deeply tied to Lean verification. Wolfram Notebook Assistant can turn conversational input into executable Wolfram Language and supports notebook chat / programmatic LLM use.
Lean verifier + mathlib	Yes	Lean is a real proof assistant, and mathlib is a large active formalized mathematics library, but it is not automatically connected to the informal literature.
AI proof assistant inside Lean	Partly	Lean Copilot, LeanDojo, LeanSearch, Loogle, REPL/Pantograph, etc. exist, but they are tools, not a full research cockpit. Lean Copilot lets LLMs suggest tactics, premises, and proof searches inside Lean.
Jupyter/Lean notebooks	Partly	`lean4-jupyter` exists and can execute Lean commands, theorems, tactics, and import Mathlib, but it is not the dominant Lean research workflow.
CAS ↔ Lean bridge	Research prototype	There has been work on a Lean–Mathematica interface, where Mathematica computations can be used and then verified in Lean, but this is not a mainstream product-level integration.
arXiv / MathSciNet / literature graph	Partly	MathSciNet has curated citation data; zbMATH Open has networked math literature data, arXiv math preprints, and APIs; Semantic Scholar and OpenAlex provide broad scholarly graph APIs. But these are mostly bibliographic/citation graphs, not verified theorem-dependency graphs.
Statement-level math search	Emerging	Matlas, announced on arXiv in April 2026, is directly aimed at semantic search over mathematical statements, extracting 8.07 million statements from 435,000 peer-reviewed papers plus 1,900 textbooks. That is close to one missing layer, but not the whole platform.
Simulator	Domain-specific	Mathematica, SageMath, Python, Julia, etc. can simulate, but “the simulator” depends on the mathematical domain. SageMath is an open-source mathematical software system built on NumPy, SciPy, SymPy, Maxima, GAP, FLINT, R, and more.
Tutor	Partly	Wolfram has AI tutoring/course-assistant features, and general LLMs can tutor, but no trusted tutor that is simultaneously literature-aware, theorem-aware, Lean-aware, and computation-aware.

So the honest answer is:

The ingredients exist. The integrated platform does not.

And yes, that is surprising.

Why it does not exist yet

The hardest problem is not UI. The hard problem is semantic alignment.

Informal mathematics lives in PDFs, LaTeX, notation conventions, hidden assumptions, diagrams, examples, folklore, “it is clear that,” and decades of context. Lean lives in fully explicit definitions, types, imports, namespaces, hypotheses, and proof terms. Mathematica lives in symbolic/numeric computation with its own assumptions and simplifications. Literature graphs live mostly at the paper/citation/author level. These four worlds do not share one ontology.

A platform like the one you describe would need to translate between:

informal theorem in a paper
↔ formal Lean statement
↔ computational experiment
↔ relevant literature graph
↔ human explanation
↔ verified proof object

That translation is still fragile.

The closest existing “stacks”

The closest commercial-feeling stack is probably:

Wolfram Notebook Assistant + Mathematica + Wolfram Language + LLM Kit

That gives you computation, notebooks, natural-language control, code generation, visualization, and some tutoring. But it does not give you deep Lean/mathlib verification or literature-graph-grounded research navigation.

The closest formal-math stack is:

Lean + mathlib + LeanSearch/Loogle + Lean Copilot + LeanDojo + VS Code

That gives you verification, formal library search, AI-assisted proving, and machine interaction. But it does not give you Mathematica-level computation, arXiv/MathSciNet-style semantic navigation, rich notebooks, simulation, and tutoring as one product.

The closest literature/intelligence stack is:

Semantic Scholar / OpenAlex / zbMATH Open / MathSciNet / Matlas

That gives you discovery, citation graphs, bibliographic metadata, and now some statement-level semantic search. But it does not automatically turn the literature into formal definitions, Lean theorem statements, verified dependencies, or executable simulations.

What is actually missing

The missing thing is a mathematical knowledge compiler.

Not just a chatbot. Not just a proof assistant. Not just a CAS. A compiler from messy human mathematics into a structured workspace:

Ingest papers/books/notes.
Extract definitions, theorems, examples, dependencies.
Build a concept graph, not merely a citation graph.
Match informal statements to existing mathlib objects.
Generate Lean skeletons.
Run computational experiments in Mathematica/Sage/Python.
Ask an AI agent to propose conjectures/proofs/counterexamples.
Verify what can be verified in Lean.
Explain everything back to the human.
Store the result as reusable mathematical infrastructure.

That platform would be huge.

The reason it feels inevitable

Because every component is moving toward it.

Lean is becoming a serious platform for formal mathematics. Wolfram is adding LLM-native computational notebooks. Semantic Scholar/OpenAlex/zbMATH expose literature data. Matlas is attacking semantic search over mathematical statements. Lean Copilot and AlphaProof-like systems show that AI agents can operate inside formal proof environments.

So the platform you described is not science fiction. It is more like an obvious missing integration layer.

My take

There is no “Mathematica + Lean + arXiv graph + notebook + AI agent + simulator + tutor + verifier” platform yet because the field has been split into tribes:

CAS people build computational power.
Proof-assistant people build verified foundations.
ML people build agents.
Library/database people build literature graphs.
Educators build tutors.
Mathematicians still mostly live in LaTeX and PDF.

The opportunity is to fuse them.

The first successful version probably will not try to cover all of mathematics. It will pick a narrow domain — say undergraduate real analysis, graph theory, algebra, probability, or olympiad geometry — and build the full loop there:

read literature → explain concepts → run examples → suggest conjectures → formalize statements → verify proofs → teach the user.

That would already be revolutionary.

Game Studio

Fri, 08 May 2026 07:19:51 GMT

I think we can build really great very lean game studio and produce really hit games!

cachehit

starting with roblox games and getting the money first

https://github.com/MaximumADHD https://github.com/pizzaboxer

https://rojo.space/

fertilizer plant

Fri, 08 May 2026 07:19:51 GMT

We import lot of fertilizer from outside of the country, I think we can vertically integrate and provide the supply!

https://youtu.be/KGAPm_2GoLM?si=AmQ2pgW_bFi8PH72

Need to solve gas import

Industrial Gases

Oxygen, nitrogen, argon, hydrogen — used in hospitals, steel, food packaging, chip manufacturing
Made by maybe 4-5 companies globally
Air Liquide, Linde operate like quiet monopolies

this is the real main game changer: https://bindwell.ai/

neural computer

Fri, 08 May 2026 07:19:51 GMT

https://flipbook.page/

rare earth Magnets

Fri, 08 May 2026 07:19:51 GMT

China is dominating by controlling 90% of rare earth minerals and making all people kneel from robotics motors to lot of other things, like rare earth magnets are one of the hugs part.

Government need to liberaliz and build public-private relation with separation of concern and responsibility.

Rare earth magnets market is going to dominate in the near future. the pie is so huge that it will never be crowded, there always be place to serve the demand.

what is their method how they are doing this: https://www.nironmagnetics.com/

Rare-Earth-Free Magnets

Revenue Potential: $15B by 2035 (market-creating if successful)
Problem: NdFeB requires REEs. Ferrite is too weak for EVs. Alternatives are lab-scale:
- MnBi (Manganese-Bismuth): Ames Lab working on it; complex synthesis
- α”-Fe16N2 (Iron Nitride): Niron Magnetics (backed by GM, DOD, Stellantis) — closest to commercial but still pre-scale; metastable phase decomposes above 539K
- High-Entropy Borides: Georgetown University breakthrough — earth-abundant transition metals + boron, rare-earth-free AND precious-metal-free
- MnAlC: PASSENGER project (Europe) achieved first industrial-scale 158kg production; validated in water pump and e-scooter prototypes
Breakthrough Needed: Stabilize Fe16N2 phase at industrial temperatures; scale MnAlC to tonnes; AI-driven materials discovery.
Risk: Extreme. But if it hits, it leapfrogs the entire rare earth supply chain.
Timeline: 10+ years

self balancing music player

Fri, 08 May 2026 07:19:51 GMT

I want to make my own music player which self balances itself with cool rgb cube and projection and when i move in certain things it should interactively change the music and other things maybe a projector or something with alexa kind of system which is built in itself.

https://willempennings.nl/balancing-cube/ https://github.com/willem-pennings/balancing-cube/ https://oshwhub.com/nokia82/cubli_-ping-heng-li-fang-ti

this guy has really great UI and display ideas: https://x.com/adamfuhrer https://www.instagram.com/adamfuhrerart and mohenjo daro style art would do just fucking great

bare metal shaders would just do crazy thing

https://www.olwal.com/ has good idea about HCI

holograhpic diplay or something!

and vinyle record player would also be cool idea

Tilt to control — nudge it and it recovers, but a deliberate tilt in a direction skips tracks, changes volume, pauses. The physics is the UI.
Spin to shuffle — give it a twist, the reaction wheels absorb it, shuffle mode activates.
Pick it up to pause, set it down and it jumps back up and resumes.
The “holographic” display could be a pepper’s ghost prism on top (cheap, no special tech) or a small transparent OLED — both work while spinning/tilting.

https://www.reddit.com/r/IndustrialDesign/comments/1sfo2fl/snapshots_inner_workings_of_my_cube_lamp_design/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

attach cool display to show current playing music song:

https://github.com/remrc/Self-Balancing-Cube

base plate is wireless charging thing, just drop it and it would charge

stoned monk engineering

Tier	Material	Machining Difficulty	Margin
Standard	Aluminum	Easy	Best
Stealth	Anodized black aluminum	Easy	Best
Heritage	Brass	Medium	Great
Aerospace	Titanium	Hard but predictable	Good
Obsidian	Zirconium (heat-blackened)	Hard but predictable	Good

stockeeda

Fri, 08 May 2026 07:19:51 GMT

Sexy Mobile App which just works http://everyticker.com/ (thiel fellow) with geo politics tracking and custom dashboard something like monitor.stockeeda.com https://github.com/koala73/worldmonitor maybe some way if we can figure out way to get aggregated portfolio tracking across the broking app then community and leaderboard in close friends group and across the app

interesting use case

video semantic meaning for security teams

Fri, 08 May 2026 07:19:51 GMT

basically semantic search over security team feeds and with that geo world model we can push things to the next level, and from employee level tracking to productivity to summary and everything this thing can become game changer

traditional CCTV camera feed with context to get info and meaning in crazy level

Tech is very small and solvable problem here, brand, distribution and how to scale this thing greatly is the thing which need to think about!

writing

Fri, 08 May 2026 07:19:51 GMT

w Ancient / Classical

Marcus Aurelius — Meditations (180 AD)
Seneca — Letters from a Stoic

Renaissance / Early Modern

Montaigne — Essays (1580) — literally invented the essay form
Francis Bacon — Essays (1597)

18th Century

Benjamin Franklin — Autobiography (1791)
Samuel Johnson — The Rambler essays

19th Century

Ralph Waldo Emerson — Self-Reliance (1841)
Henry David Thoreau — Walden (1854)

20th Century

George Orwell — Politics and the English Language (1946) — read this first, before anything else
E.B. White — Once More to the Lake (1941)
James Baldwin — Notes of a Native Son (1955)
Joan Didion — Slouching Towards Bethlehem (1968)
David Foster Wallace — Consider the Lobster (2004)

Contemporary

Paul Graham — essays 2000s–present
Christopher Hitchens — Arguably (2011)

George Orwell Read Politics and the English Language first

https://mathshistory.st-andrews.ac.uk/Extras/Keynes_Newton/

This Is Water David Foster Wallace Commencement Speech https://www.youtube.com/watch?v=DCbGM4mqEVw

Christopher Hitchens His essays collected in Arguably. Teaches confidence of voice — he never hedges when he doesn’t need to. Good antidote if your writing tends to be timid.

James Baldwin Notes of a Native Son and The Fire Next Time

Michel de Montaigne https://hyperessays.net/essays/toc/

Daily Routine

Sun, 12 Apr 2026 00:00:00 GMT

The System

No decisions. Follow the clock. Jobs to be done.

Sleep: 11 PM lights out. Phone in another room from 10:30 PM. Wake: 6:30 AM. No snooze. Feet on floor.

Time	Block	What
6:30 AM - 7:00 AM	Wake + Prep	Bathroom, water, gym clothes. No phone.
7:00 AM - 8:00 AM	Gym	Lift. No thinking. Just go.
8:00 AM - 8:30 AM	Shower + Food	Simple breakfast. Still no phone.
8:30 AM - 9:00 AM	Morning Pages	30 min writing. Stream of consciousness. Pen/paper or Obsidian.
9:00 AM - 1:00 PM	Main Quest (Deep Work 1)	4 hours undivided focus. Phone in another room. ONE thing only.

Saturday AM Override

Time	Block	What
6:30 AM - 8:30 AM	Gym + Shower	Same as weekday.
8:30 AM - 9:00 AM	Morning Pages	Same as weekday.
9:00 AM - 12:00 PM	Weekly Review	3 hours. Review what got done. Pick next week’s main quest focus. Prune ideas. Plan, don’t drift.
12:00 PM - 1:00 PM	Publish	Take something from the week’s daily writing, polish it, ship it. Blog post, substack, tweet thread — something goes out every Saturday.

Sunday Override

Time	Block	What
Wake up by 8 AM	No main quest	Sleep in a bit. No deep work.
Morning	Skateboarding	Practice session. Streets, park, wherever.
Rest of day	Free	Cook, explore, adventure list, rest. No guilt.

Time	Block	What
1:00 PM - 2:00 PM	Lunch + Drift + Socials	Eat. TMKOC. Check Twitter/LinkedIn/Reddit. This is your first social media window. Scroll, reply, post — get it out of your system.
2:00 PM - 4:30 PM	Main Quest (Deep Work 2)	2.5 hours. Same rules.
4:30 PM - 5:00 PM	Break	Walk, snack, breathe.
5:00 PM - 6:30 PM	Axe Sharpening	Feeds the main quest but isn’t the main quest. Reading Skiena, Nand2Tetris, Orwell, FPGA research — whatever sharpens the current blade.
6:30 PM - 7:30 PM	Dinner + Family	Chit chat with mom. Khari toast. Protect this.
7:30 PM - 8:30 PM	Tinker Hour	Mon/Wed/Fri = Guitar (30 min timer, sit with it). Tue/Thu/Sat = Hardware (cube, FPGA, physical stuff). Sunday = Free.
8:30 PM - 9:30 PM	Reading	Fiction or essays. Not tech. Orwell, Montaigne, Baldwin, Didion. This expands thinking.
9:30 PM - 10:00 PM	Write	One paragraph minimum. Blog draft, substack post, notes on reading. Doesn’t have to be good.
10:00 PM - 10:30 PM	Socials (Window 2)	Second and last scroll of the day. Post if you have something. Then phone goes to another room.
10:30 PM - 11:00 PM	Wind Down	No screens. Prep for next day — lay out gym clothes, decide tomorrow’s main quest task.
11:00 PM	Sleep

The Rules

No phone before 9 AM. Kills the doomscroll-on-waking loop.
One main quest at a time. Pick ONE project. Minimum 2 weeks before switching.
6.5 hours of deep work is a lot. Protect those hours, waste the rest guilt-free.
Sunday is different. Sleep in till 8. No main quest. Explore, cook, adventure, rest.
Track with a paper checklist. Date + blocks. Check them off. That’s it.
Never miss twice in a row. Skip gym Monday? Fine. Tuesday is non-negotiable. One bad day stays one bad day.
Serendipity through doing, not reaching out. Do the work, post about it online. Let people come to you. The social media windows are for sharing what you built/learned/read — not cold DMs.
Skateboard is commute. Use it to get around during the day. Sunday is for practice.

Tinker Hour Rotation

Day	Activity
Monday	Guitar
Tuesday	Hardware
Wednesday	Guitar
Thursday	Hardware
Friday	Guitar
Saturday	Hardware
Sunday	Free

2025-2026

Sat, 11 Apr 2026 19:52:55 GMT

So this is rough plan, i am not sure whether i will be in bangalore or in ahemdabad or where at that time, but idea is to complete them as much as possible, cause all things i want them to happen!

Every project needs to be somehow contributing to community or other new persons to learn more, fun, and something which i learn a lot and enjoy with good twist, and also idea is to deploy and cary out as many experiments as we can at scale so we can have more information about things, we never know what may work or what not, and what good things can come out of any experiment.

maximising surface area of luck and doing everything in order to achieve it.

projects:

Books:

Skills:

mastering them

rust
python
pytorch
jax
assembly
maths
reverse engineering
skateboarding
video editing
electric guitar
cooking
writing essays and thoughts more on substack(ability to express my thoughts more easily)

goals/metrics:

companies on the radar(grab opportunity if given or create):

Nvidia recommendation engine team(https://developer.nvidia.com/blog/author/eoldridge/)
world models at Meta(Yann LeCun)
Anthropic(Mech Intr)
Deepmind(Mech Intr)

http://neuralnetworksanddeeplearning.com/

Movies:

“Stories We Tell” (2012) directed by Sarah Polley:
- Context: A documentary that investigates the director’s own family secrets. It creates a “meta-narrative” about how different people remember the same event differently, directly paralleling the themes of truth and fabrication in F for Fake.
“Close-Up” (1990) directed by Abbas Kiarostami:
- Context: Based on a true story about a man who impersonated a famous film director. The movie features the actual people involved re-enacting the events. It is a masterpiece of deconstructing the “truth” of cinema.
“The Gleaners and I” (2000) directed by Agnès Varda:
- Context: Varda is the godmother of the cinematic essay. This film is a wandering, personal exploration of “gleaning” (collecting leftovers), but it is structured by her own curiosity rather than a traditional plot, much like a video essay.
“Exit Through the Gift Shop” (2010) directed by Banksy:
- Context: Startlingly similar to F for Fake. It starts as a documentary about street art but evolves into a question of who is conning who. It questions the value of art and the “expert” opinion.

Adventure & Explore

Take a train to a random city and figure it out when you get there
Sleep under the stars somewhere you’ve never been
Visit a town with a population under 500
Eat alone at a fancy restaurant and enjoy every second of it
Get on a flight with only a carry on and no hotel booked
Find the highest point in your city and watch the sunset from there
Walk into a museum you’ve never noticed before
Spend a full day in a neighborhood you always drive past

Strike up a real conversation with a stranger at a café and actually listen
Write a letter to someone who changed your life and actually send it
Attend an event where you know absolutely no one
Compliment five strangers in one day and mean every word
Call an old friend you haven’t spoken to in years with no agenda
Cook a full meal for someone who needs it right now
Ask an elderly person about their biggest regret and listen without your phone

functional programming

Sat, 11 Apr 2026 19:52:55 GMT

https://people.csail.mit.edu/rachit/post/why-study-programming-languages/

https://steshaw.org/plt/

https://cs.lmu.edu/~ray/notes/languagetheory/

https://mpc.sh/blog/a-gentle-intro-to-plt

https://en.wikibooks.org/wiki/Introduction_to_Programming_Languages

hardcaml FPGA

Sat, 11 Apr 2026 19:52:55 GMT

hardcaml hardcaml custom open source FPGA SSD with hardcaml FPGA book on hitchikers guide to FPGA

all this things: https://intel.github.io/rohd-website/

open source FPGA silicon : https://github.com/MidstallSoftware/aegis

https://anil.recoil.org/notes/fpgas-hardcaml https://anil.recoil.org/ideas/tracing-hdl-with-effects https://anil.recoil.org/ideas/computational-storage-for-vector-dbs

claude conversation:

what will we achive which hardcaml can’t do with this thing or in genreal comparsion with other hdl in the ecoyssytem from dart to haskell clash

https://anil.recoil.org/ideas/tracing-hdl-with-effects

Great read — this is a research proposal by Anil Madhavapeddy (Cambridge) for an MPhil project exploring OCaml 5 effects as a foundation for a tracing-style HDL. Let me break down exactly what this achieves vs. HardCaml, and how it positions in the broader ecosystem.

The Core Idea: Tracing vs. Building

The fundamental distinction is how the circuit graph gets constructed.

HardCaml today — combinator / data-structure style: You write OCaml that builds up a circuit as a data structure. Your OCaml program’s output is the circuit. This requires heavy use of the OCaml module system (functors, first-class modules, typed interfaces) to compose components. It’s powerful but has a steep learning curve — you’re always mentally managing two layers: the OCaml meta-level and the hardware object-level.

The effects-based tracing HDL — direct execution style: You write OCaml that directly looks like the circuit it describes. You evaluate OCaml code, and the effect handlers intercept and record what happened — building the circuit graph as a side effect of running the program. The circuit emerges from tracing execution, not from explicitly constructing a data structure.

What This Buys Over HardCaml Specifically

1. Ergonomics & accessibility HardCaml’s module-system-heavy approach is a productivity wall. It makes very heavy use of the module system to build up the circuit description as an OCaml data structure recoil, which means you need to be fluent in OCaml’s more advanced features (functors, module signatures, first-class modules) before you can be productive. The effects approach lets you write code that reads like ordinary imperative/functional computation — the tracing machinery is transparent.

2. Control-flow memoization via effects This is the deeper technical win. Effect handlers can intercept recursive calls and memoize their circuit-graph expansions, which means shared subgraphs don’t get duplicated. In a plain combinator HDL you have to manually thread sharing; here the handler does it structurally.

3. Restricting recursion properly Arbitrary recursion in hardware is a problem (it implies infinite logic). Effects let you intercept recursive calls and decide how to handle them — either unrolling a fixed number of times, requiring explicit register boundaries, or rejecting unbounded recursion statically. This is a semantic guarantee you can enforce via the handler rather than relying on programmer discipline.

4. Fixed-length vectors ergonomically Hardware is full of [N]bit types. HardCaml encodes these but it’s verbose. The proposal explicitly targets this as an ergonomics win to explore.

Comparison Across the Ecosystem

HDL	Host language	Style	Key strength	Key weakness
Verilog / VHDL	Domain-specific	Structural/RTL	Industry standard, tool support	Weak abstraction, verbose, no types
HardCaml	OCaml	Combinator / data-structure	Strong typing, Jane Street production use, module polymorphism	Module-system learning cliff, two-layer mental model
Clash	Haskell	Tracing (type-directed)	Looks like Haskell functions, REPL, fixed-length Vec types, type inference	GHC type-level complexity, long compile times, Haskell ecosystem barrier
Chisel	Scala	Embedded DSL / combinator	JVM ecosystem, industry adoption, good tooling	Scala complexity, Spark-style boilerplate, runtime errors
SpinalHDL	Scala	Combinator	Cleaner than Chisel, good abstractions	Niche, small community
Amaranth (nMigen)	Python	Combinator	Accessible, fast iteration, Python ecosystem	Dynamic typing catches errors late, Python perf for large designs
Lava / Kansas Lava	Haskell	Combinator (arrows)	Theoretically clean	Effectively abandoned
This proposal (OCaml effects HDL)	OCaml 5	Tracing via effects	See below	Research prototype — not production

The Clash Comparison Is Central

Clash uses a number of advanced Haskell type-level features to encode fixed-length vectors and has an interactive REPL that allows for exploration without requiring a separate test bench. recoil

Clash is the closest spiritual ancestor. It achieves the “write hardware like functions” feel through Haskell’s lazy evaluation + GHC’s rewriting + type-level naturals for Vec n a. The tracing HDL proposal wants to achieve the same feel in OCaml, but through a structurally different mechanism:

Clash traces by exploiting Haskell’s lazy graph reduction — the program graph is the circuit graph, extracted by GHC’s normalizer. You’re fighting GHC internals.
Effects HDL traces by using explicit delimited continuations (effect handlers) to intercept evaluation. This is semantically cleaner, more inspectable, and doesn’t depend on compiler internals.

The OCaml approach could be more predictable than Clash because you don’t rely on GHC’s optimizer doing the right thing — the tracing boundary is explicitly defined by the effect handler.

The Dart Angle

There’s no serious Dart HDL in the ecosystem. The closest thing would be using Dart for FPGA scripting/tooling (e.g. ROHD from Intel uses Dart as the host), which is more in the Amaranth territory — accessible dynamic language, combinator style. It gains Python-like ergonomics but loses the type-level guarantees that make Clash/HardCaml interesting. The effects proposal is at the opposite end: it wants more semantic power and correctness guarantees, not less.

What Would Genuinely Be Novel

If the effects approach works, the real contribution is:

A new point in the design space: tracing-style HDL that doesn’t require a lazy language or compiler internals to work, but instead uses first-class algebraic effects as the mechanism.
Effect handlers as a hardware/software duality tool: the same OCaml code could be run natively for simulation (handler = execute normally), or traced for synthesis (handler = build circuit graph), or partially evaluated (handler = mixed). One program, multiple interpretations — this is the “finally tagless” / multi-stage dream done cleanly.
Recursion control as a first-class citizen, not a linting afterthought.

The open question is whether OCaml 5 effects are expressive enough to handle the multi-shot continuation patterns needed for circuit sharing and unrolling. The memo.ml example in the effects-examples repo suggests yes, but at scale in hardware design it remains to be proven.

what would be good project to implment in this hdl to get idea about how good ocaml5 effects are there, like open source fpga or some other thing or suggest something good

This is a great question. You want something that stresses the effect system in interesting ways — not just “does it work” but “where does it shine or crack.” Let me think through this carefully.

What Makes a Good Test Case

You want a project that exercises:

Structural sharing — same subgraph reused many times (tests memoization via effects)
Controlled recursion — natural recursion that must be unrolled to fixed depth
Fixed-length vector arithmetic — the ergonomics win over HardCaml
Multiple interpretations — same code simulated, synthesized, and maybe formally verified

Tier 1: Best Bang for Effort

A Systolic Array for Matrix Multiply

This is the perfect first serious project. A systolic array is a grid of identical processing elements (PEs) where data flows rhythmically between neighbors.

Why it’s ideal:

The PE is defined once, instantiated N×M times — directly tests whether effect-based sharing deduplicates the graph properly or explodes it
The grid wiring is naturally expressed as a recursive function over indices — tests controlled recursion unrolling
It’s the exact architecture inside TPUs, and is a hot target for open-source FPGA ML accelerators (like FINN)
Small enough to complete, deep enough to be real

What you’d learn about OCaml effects: whether memoization via effects correctly identifies structurally identical PE instances, and whether the tracing produces a flat efficient netlist rather than N×M duplicate subgraphs.

A Pipelined RISC-V Core (RV32I)

Not a full core — just the 5-stage pipeline: Fetch → Decode → Execute → Memory → Writeback.

Why it’s ideal:

Pipeline hazard logic is naturally expressed as conditional forwarding — great test of how control flow in the host OCaml maps to mux logic in the traced circuit
Register file is a fixed-length vector of 32 registers — directly tests the Vec ergonomics
There are excellent open-source reference implementations (PicoRV32, VexRiscv, SERV) to validate against
The RISC-V spec is clean and minimal — you spend time on the HDL problem, not architecture archaeology

What you’d learn: how well recursive/structured decode logic traces, and whether the effect handler correctly handles the stateful pipeline registers vs. combinational logic distinction.

Tier 2: Sharper Tests of Specific Effect Properties

A Parametric FIR Filter Bank

A bank of N FIR filters sharing tap coefficients.

Why interesting: this is the canonical test for structural sharing. In a naive combinator HDL you’d get N×M multipliers. With correct memoization effects, shared constant-coefficient multipliers should collapse. This is a very targeted stress test of the memo effect specifically — you’d know within 200 lines whether the approach works or has fundamental issues.

A SHA-256 / AES-128 Core

Cryptographic cores are beloved FPGA targets and have:

Highly regular round structures (natural recursion to unroll)
Fixed 32/64-bit word widths everywhere (Vec ergonomics)
Well-known reference outputs for testing

The round function is defined once and applied N times — a direct test of whether for-style unrolling via effects produces correct pipelined or iterative hardware. There’s also an open-source reference in every HDL imaginable to compare netlists against.

A UART / SPI Controller

Deceptively good for testing the simulation vs. synthesis duality. A UART is a simple state machine, but:

You want to run it in OCaml natively with a mock byte stream (simulation handler)
Then trace it to Verilog (synthesis handler)
Then maybe hook it to a formal tool to verify the baud rate arithmetic (verification handler)

Same OCaml code, three handlers, three interpretations. This is the cleanest possible demo of the effects-as-multiple-backends idea and fits in ~500 lines.

Tier 3: Ambitious / Research-Grade

A Sparse Attention Accelerator

Given the FPGA+vector-DB usecase Anil mentions on the same site, implementing a hardware block for sparse dot-product attention would be genuinely novel. The challenge: irregular sparsity patterns mean the circuit structure itself becomes data-dependent — a hard case for any tracing HDL.

Connecting to OpenROAD / Yosys

Rather than targeting Verilog output → vendor tools, you could target the RTLIL (internal format of Yosys, the open-source synthesis tool) directly from your effect-traced AST. This would make the whole flow open: OCaml effects → RTLIL → OpenROAD → GDS. No proprietary tools anywhere.

Concrete Recommendation

Start with the FIR filter bank (2–3 days, laser-focused on the memo effect), then graduate to the systolic array (the real project). These two together will tell you definitively whether OCaml 5 effects are genuinely better than Clash’s lazy-graph approach or HardCaml’s explicit modules — specifically on the sharing and recursion problems that matter most in real synthesis.

The UART is the right companion project to validate the multi-handler / multi-interpretation story, which is arguably the most novel claim in the whole proposal.

kids metal songs

Sat, 11 Apr 2026 19:52:55 GMT

this is good genre music, and vibe is also good, i really liked it, i think this and cool beats and music and rap would be killer combo

GC-Fuzz

Wed, 08 Apr 2026 18:39:54 GMT

https://dl.acm.org/doi/epdf/10.1145/3733598 https://github.com/BurryPotter/GCFuzz https://github.com/JikesRVM/JikesRVM

read this shit first: https://bushido-sec.com/index.php/2025/01/03/fuzzing-harness-guide/

paper: https://rmod-files.lille.inria.fr/Team/Texts/Papers/Poli23a-ICST-HeapFuzzing.pdf

start with fuzzing mmtk: https://github.com/mmtk/mmtk-core

old harness: https://github.com/JikesRVM/JikesRVM/tree/master/MMTk/harness

https://github.com/rust-fuzz

https://rust-fuzz.github.io/book/introduction.html

https://www.steveblackburn.org/research/projects/gc-fuzzing/

once harness is done unleash the llms: https://arxiv.org/html/2512.03420v2

things i have gone throug:

https://tutorial.ponylang.io/gotchas/garbage-collection.html#garbage-collection-in-pony https://tutorial.ponylang.io/runtime-basics/garbage-collection#actor-collection-inter-actor-gc-via-orca https://www.mono-project.com/docs/advanced/garbage-collector/sgen/working-with-sgen/ https://www.mono-project.com/docs/advanced/garbage-collector/sgen/

https://www.youtube.com/live/lGM37Z3oDKw?t=12105s

Rex-Minecraft Robo Dog

Wed, 08 Apr 2026 18:39:54 GMT

Rex is my Minecraft dog, so i am bringing to this real life

https://www.reddit.com/r/robotics/comments/1rqvaeu/deep_robotics_has_built_a_robot_horse_seemingly_a/

make it unicorn whih can fly also

link: https://www.instagram.com/p/DMdc7UyIyDc/?utm_source=ig_web_copy_link

Component	Quantity	Price per Unit (INR)	Total (INR)
Servos (LX-16A)	12	750	9,000
Jetson Orin Nano	1	45,000	45,000
Battery (2S LiPo)	1	1,500	1,500
Sensors (Camera + IMU)	1 + 1	1,000 + 100	1,100
Audio Devices (Mic + Speaker)	1 + 1	500 + 500	1,000
Power Management (Regulator)	1	200	200
3D Printing (Chassis)	1	5,000	5,000
Total			63,800

Component Category	Item	Estimated Unit Cost (USD)	Quantity	Estimated Total Cost (USD)	Estimated Total Cost (INR)	Notes
Main Processing Unit	NVIDIA Jetson Orin Nano Super Developer Kit	$249.00	1	$249.00	₹21,508.62	Chosen for its AI performance and ecosystem.
Actuators (Servos)	High-Torque Digital Serial Bus Servo (e.g., Hiwonder HX-35H)	$25.00 (average)	14	$350.00	₹30,233.00	12 for legs (3 per leg), 2 for head/tail.
Servo Control	ROS Robot Control Board with STM32	$100.00 (estimate)	1	$100.00	₹8,638.00	For managing multiple servos and ROS integration.
Power Management	High-Capacity LiPo Battery Pack (e.g., 12V 3000mAh)	$40.00 (estimate)	1	$40.00	₹3,455.20	For robot’s operational autonomy.
	DC-DC Buck Converter	$8.00 (estimate)	1	$8.00	₹691.04	To provide stable voltage to different components.
Sensors	High-Definition Camera Module (e.g., RPi V2/V3 compatible)	$30.00 (estimate)	1	$30.00	₹2,591.40	For AI vision tasks like object recognition.
	IMU Sensor (9-axis)	$45.00	1	$45.00	₹3,887.10	For balance, orientation, and stability.
	USB Audio Codec Module with Microphone	$72.00 (estimate)	1	$72.00	₹6,219.36	For natural language input to the LLM.
	Small Speaker Module	$5.00 (estimate)	1	$5.00	₹431.90	For audio output and robot responses.
Fabrication Materials	3D Printing Filament (PETG/ABS, ~3kg)	$25.00 (per kg estimate)	3	$75.00	₹6,478.50	For chassis and limb components.
Miscellaneous	Screws, nuts, wires, connectors, tools, etc.	$150.00 (estimate)	1	$150.00	₹12,957.00	Essential small parts and assembly tools.
Total Estimated Cost	---	---	---	$1,184.00	₹102,653.12	Please note: These are estimates. Prices can vary due to vendors, specific models, and import duties/taxes in India.

https://docs.kscale.dev/docs/ksim#tutorial

https://www.youtube.com/watch?v=c64FnSvj8kQ&t=166s

idea is to create cheap(<$1000) DIY minecraft style robot kit which just works and easy to setup, and moduler enough so people can do whatever they want to do!

need good amount of money for this, and need more foundation, need to apply to all fellowships

make it do skate and make it also fly fucking fly!

https://www.youtube.com/watch?v=S4eQXXxUnNE

8th April 26

Wed, 08 Apr 2026 18:39:54 GMT

It’s been like hazy morning, waking up then actually waking up after 1 hr, taking a dump with information overloading my brain, i think this is what normal day from last few weeks is looking like, consuming lot of social media and news tech, articles, blog and after sudden crash drifting in the time slowly, just thinking about what can be done or this is good idea etc.., till this happens it is around 12, then i am little hungry, then i usually find my self crawling to the food and having it while watching tmkoc, then slowly drifting away in the sleeping position while feeling i am doing some productive while the whole thing is just bs and productivity porn kind of thing, i fell to sleep, while i am sleeping or which feels like i am sleeping my mind is quite active and it’s been screaming that i should do this i should do that this is the time etc.., i every day think about going to gym in the morning and in the evening but i don’t go cause in the morning i am mostly sleeping and in the evening i am doomscrolling or doing some random shit which i don’t know feels very important in that present moment but it is always useless thing, if i would have skipped it, it wouldn’t have any single impact or change, but still i end up doing it, after i wake up i usually i goto main hall do chit chat and brain picking of my mom and enjoy khari toast and just simply drift away in the time more, then i either go to garden for a walk for 1 hr or just simply find some other intellectual curiosity think and just drift away in that, i complete my dinner same daily food ritual and doing chitchat with my mom, then i go to my room, in the whole day if i got some motivation then i play guitar A and D chord to just practice but not for more than 5min, cause those 5min feels like total 1 hr, after dinner i then again start koala ritual again drifting away either slowly in the nap or again some random rabbit hole without doing any thing and thinking we will do something tomorrow, and i mostly go to sleep around 1 or 2 AM, and the day starts again like this on the next days, same thing no difference, just few here and there minor changes.

This is what is happening from almost daily after i left the Krupal and the project i think around 2-3 weeks, in initial weeks i had projects and ideas and deadlines so i was working towards them, i really work great when i have some kind of realistic deadline ahead of me and because i don’t have deadline or some hanging sword above me last week has been just drifting the time and just feeding my mind with all kinds of information which feels like i am doing something important or making some progress but where and which direction no one knows. I have been fantasising the whole Chennai thing and piggybacking on that, which i feel is wrong thing but my whole mind is not able to stop thinking about it, the deadline for that thing is around 20th April, i almost submitted my application 1 month ago to the deadline, I know they are going to grill my really hard on the interview cause if i were them then i would have also done the same thing, cause on paper many people look really great but in real they are many times are not, and in the age of LLM it is going to get more worse then worse, so i am expecting getting really grilled by them, and that’s why the whole last 1.5 week i have been thinking about getting my self ready and prepared for the thing, but i am just drifting away from the thing i think main reason is somewhere my inner self thinks that as deadline and things move in the fast pace momentum i will wake up and lock in, but i know those last moment things one always try to sneak in and find the hacks and rushes the whole thing, but i don’t want to do that this time, cause this whole thing can really change the whole thing and future for me, and where the things are going i think that is the best opportunity for piggy backing and building great intuition and discipline.

So start doing small things and complete them like jobs to be done and forget about other things no overthinking just it is jobs to be done!

Co2 Remove

Tue, 07 Apr 2026 14:27:56 GMT

Ocean-Based Carbon Removal is genuinely underexplored and potentially massive margin territory. Two variants:

Seaweed farming at scale — grow kelp, it absorbs CO₂, sink it to deep ocean. Permanently removed. Cost potentially very low given India’s coastline. But methodology approval is still nascent — registries haven’t fully figured out how to verify it yet. Early mover advantage is enormous here.

Ocean Alkalinity Enhancement (OAE) — essentially ERW but done in seawater instead of farmland. Crush limestone or similar minerals, dissolve in ocean, accelerates natural carbon absorption. Some people think this is the largest scale carbon removal pathway that exists — oceans can absorb vastly more than land. Still very early stage scientifically but a few startups globally are working on it.

Soil Carbon Sequestration via Regenerative Agriculture — this is what Varaha mainly does alongside biochar. Lower margin ($15-50/tonne) but the volume potential is staggering given India’s farm coverage. Margin per credit is thin but if you’re doing 10 million tonnes a year the absolute numbers work.

Peatland and Wetland Restoration — one of the most carbon-dense ecosystems on the planet. Degraded peatlands release enormous CO₂. Restoring them locks it back. India has some peatland, Southeast Asia has massive reserves. Credits sell at $50-150/tonne and the science is well established. Very underpursued in India.

Mineralization via Industrial Waste — some industrial processes produce alkaline waste (steel slag, cement kiln dust, fly ash) that naturally absorbs CO₂ when exposed to air. You’re essentially getting paid to dispose of industrial waste and generating carbon credits from the same transaction. Margin structure is very interesting — negative feedstock cost essentially.

sabbatical stuff

Tue, 07 Apr 2026 14:27:56 GMT

thinking for going on the learning sabbatical vacation!

things which i am learning and completing:

Core:

Nand2Tetris
SICP
⁠⁠math book jeremy kun (https://agorism.dev/book/math/intro/Dr.%20Jeremy%20Kun%20-%20A%20Programmer%E2%80%99s%20Introduction%20to%20Mathematics-CreateSpace%20Independent%20Publishing%20Platform%20%282018%29.pdf)
An Infinitely Large Napkin (https://web.evanchen.cc/napkin.html)
Game engine architecture by jason gregory

Scaling:

Performance Engineering: https://www.youtube.com/playlist?app=desktop&list=PLUl4u3cNGP63VIBQVWguXxZZi0566y7Wf
distributed systems: https://www.youtube.com/playlist?list=PL4YhK0pT0ZhXTRSAYHAgBcJkhlM2hlhw3
competitive programing
long form essays writing
creating content in all forms(text, images, videos, reels, short, yt vlog everything)

AI:

JAX
https://jax-ml.github.io/scaling-book/
world models
RL
fast ai deep learning course

Robotics:

Music:

electric guitar
carnatic music

photography:

traveling and clicking some photos with good camera

https://github.com/skyzh/mini-lsm https://github.com/skyzh/write-you-a-vector-db

https://dlsyscourse.org

https://www.cs.cmu.edu/~mgormley/courses/10601/

https://spinningup.openai.com/en/latest/

Journey to the cool

Tue, 31 Mar 2026 06:11:28 GMT

Trazillion dollar idea.

Make a game engine, a decent one, that supports basic stuff, but that could be heavily extended and modified. But make it unconventionally console / test driven, make it easily verifiable. Then, make a lot of toy game with it. Get lot of instructions, Train (fine-tune) an LLM on it. Sell the LLM & game engine as a paired product. Train on the user-generated games. SFT. RLVR.

https://entropygames.ai/research/building-the-next-generation-of-games?v2&curius=1819

Karan's Notes

Ai's Mathmetica

Core thesis

The argument in one sentence

Main moves in the essay

His three big caveats about AI-for-math

What he thinks mathematicians should do

My read

Optimize for understanding, not just proof production

For a student

For a researcher

For someone choosing a career

What to cultivate now

The dangerous traps

The simplest answer

Why it does not exist yet

The closest existing “stacks”

What is actually missing

The reason it feels inevitable

My take

Game Studio

fertilizer plant

neural computer

rare earth Magnets

Rare-Earth-Free Magnets

self balancing music player

stockeeda

video semantic meaning for security teams

writing

Daily Routine

The System

Saturday AM Override

Sunday Override

The Rules

Tinker Hour Rotation

2025-2026

projects:

Books:

Skills:

goals/metrics:

companies on the radar(grab opportunity if given or create):

Adventure & Explore

Social & Human Connection

functional programming

hardcaml FPGA

The Core Idea: Tracing vs. Building

What This Buys Over HardCaml Specifically

Comparison Across the Ecosystem

The Clash Comparison Is Central

The Dart Angle

What Would Genuinely Be Novel

What Makes a Good Test Case

Tier 1: Best Bang for Effort

A Systolic Array for Matrix Multiply

A Pipelined RISC-V Core (RV32I)

Tier 2: Sharper Tests of Specific Effect Properties

A Parametric FIR Filter Bank

A SHA-256 / AES-128 Core

A UART / SPI Controller

Tier 3: Ambitious / Research-Grade

A Sparse Attention Accelerator

Connecting to OpenROAD / Yosys

Concrete Recommendation

kids metal songs

GC-Fuzz

Rex-Minecraft Robo Dog

8th April 26

Co2 Remove

sabbatical stuff

Journey to the cool