arthurcolle

The cloned cat may still have the opportunity to interact with the original, so it could just be taken in as a "child" kitten of the cloned cat.

The project is mostly to prove viability as I've seen focus on dogs and horses.

There is a great documentary on a scientist who is cloning Saudi Arabian horses.

I have been using a quorum composed of step-3.5-flash, Kimi k2.5 and glm-5 and I have found it outperforms opus-4.5 at a fraction of the cost

That's pretty cutting edge to me.

EDIT: It's not a swarm — it's closer to a voting system. All three models get the same prompt simultaneously via parallel API calls (OpenAI-compatible endpoints), and the system uses weighted consensus to pick a winner. Each model has a weight (e.g. step-3.5-flash=4, kimi-k2.5=3, glm-5=2) based on empirically observed reliability.

The flow looks like:

  1. User query comes in
  2. All 3 models (+ optionally a local model like qwen3-abliterated:8b) get called in parallel
  3. Responses come back in ~2-5s typically
  4. The system filters out refusals and empty responses
  5. Weighted voting picks the winner — if models agree on tool use (e.g. "fetch this URL"), that action executes
  6. For text responses, it can also synthesize across multiple candidates

The key insight is that cheap models in consensus are more reliable than a single expensive model. Any one of these models alone hallucinates or refuses more than the quorum does collectively. The refusal filtering is especially useful — if one model over-refuses, the others compensate.

Tooling: it's a single Python agent (~5200 lines) with protocol-based tool dispatch — 110+ operations covering filesystem, git, web fetching, code analysis, media processing, a RAG knowledge base, etc. The quorum sits in front of the LLM decision layer, so the agent autonomously picks tools and chains actions. Purpose is general — coding, research, data analysis, whatever. I won't include it for length but I just kicked off a prompt to get some info on the recent Trump tariff Supreme Court decision: it fetched stock data from Benzinga/Google Finance, then researched the SCOTUS tariff ruling across AP, CNN, Politico, The Hill, and CNBC, all orchestrated by the quorum picking which URLs to fetch and synthesizing the results, continuing until something like 45 URLs were fully processed. Output was longer than a typical single chatbot response, because you get all the non-determinism from what the models actually ended up doing in the long-running execution, and then it needs to get consensus, which means all of the responses get at least one or N additional passes across the other models to get to that consensus.

  Cost-wise, these three models are all either free-tier or pennies per million tokens. The entire session above (dozens of quorum rounds, multiple web fetches) cost less than a single Opus prompt.

> Its also worth noting that if you can create a business with an LLM, so can everyone else.

False. Anyone can learn about index ETFs and still yolo into 3DTE options and promptly get variation margined out of existence.

Discipline and contextual reasoning in humans is not dependent on the tools they are using, and I think the take is completely and definitively wrong.

Probably

Arbitrage!

  Location: Washington, D.C. (open to relocation, but preference to initial remote period)
  Remote: Yes
  Willing to relocate: Yes (June 2026 and beyond)
  Technologies: Python, Rust, Elixir | Neo4j, Postgres, Redis, K8s | PyTorch, MLX, LoRA/QLoRA, Modal A100s, H100s | FastAPI, Phoenix, Rails
  Résumé/CV: https://linkedin.com/in/arthurcolle
  Email: arthur@distributed.systems

Background:

Deep knowledge of OpenAI, Anthropic + LLM APIs. Able to run models on GPUs and Apple Silicon unified memory. I have trained + deployed agentic coding, tool calling and multi-turn agent models in two prior roles.

Building autonomous agent infrastructure. Currently run a small R&D shop (Distributed, Systems & Co.) doing distributed AGI and AI safety research. I am developing graphsub (graph substrate), a graph-native service mesh for agents, eval harnesses, self-modifying DSLs.

Before that: 4 years at Goldman Sachs deep in SecDB/Slang (the reactive dependency-graph system). I worked on collateralized mortgage obligations, on the interest rate products trading desk (as an engineer in the technology division), automated parts of the downstream IR-swap lifecycle and on CDS lifecycle management for novations and other lifecycle events. Had a hand in managing and working on our internal TriOptima notional compression cycles. I got promoted from tech into structuring. Series 7.

Recent work: 15-service LLM stack (zero downtime across 98 deploys), fine-tuned Llama-3.1-8B (23% hallucination reduction), Rust daemons and TUIs. 3k+ GitHub contributions last year.

OSS: agents.erl (BEAM-native cognition), https://github.com/arthurcolle/openai-mcp, fortitude (a framework for bootstrapping agents), https://github.com/arthurcolle/dspy.ex

Looking for: AI labs, defense tech, or startups building agentic systems / foundation model infra / AI safety tooling. IC or research roles.

I can run SFT and RL experiments, train models, and I'd love to be able to train agentic models and ideally with large GPU clusters. I have enjoyed these last two years, a focused period of building "agentic scaffolding" and scaling up multi-agent systems and if any of this sounds appealing to you, I can definitely be of some help to your org!

GitHub: https://github.com/arthurcolle

recent research publications:

https://apartresearch.com/project/detecting-piecewise-cyber-...

I can't seem to play the audio on iOS 26

Is this a regression?

US MIC are already planning on integrating fucking Grok into military systems. No comment.

This was my first thought as well. Prism is a cool name, but I'd never ever use it for a technical product after those leaks, ever.

Did they really vibe code a partial implementation and blog about it?

That's one way to destroy the CF blog credibility!

I mean, once we're able to run and operate multinational corporations off-world, GDP becomes something very different indeed

:(

If you would like some help with productionizing this, or help with Git or whatever, feel free to drop me an email! my user at gmail works :)

Love seeing cool, useful apps like this. Thanks for sharing!

how would that even work with https://www.reddit.com/r/nin/

Thank you so much for sharing this excellent resource.

I'm watching this youtube video (https://www.youtube.com/watch?v=tlD0C5CrWcA) and he apparently built this out to support his Master's research at the University of Maryland, where I went to school for undergrad.

The long arm of the DMV no-fly zone is no joke!

Can you discuss a bit more of the architecture?

Since them I've acquired a 3D printer, so I've increased the surface area of expensive things I can break.

If I can ever figure out how to repurpose some of these electronics maybe for some kind of AI robot (yes, the gimbal + camera optics are so nice, it feels like a sci fi eyeball from 2037!) I will be back in business.

Some people sell exploits to "jailbreaks DJI drone firmware but with current US admin I don't think it is prudent to do too much "off-label" usage of this kind of tech.

But seeing this geofencing post.... I just had too much experience trying to get around these restrictions to actually believe that they'd drop the geofencing, especially after a consumer drone ban.

Thank you for the chain suggestion, that would have been intelligent to do. Matter of fact, my father may have made that suggestion at the time. Alas, that was a very move fast and break things period of my life.

Feel free to reach out, no worries :)

I was unable to ever get it to fly reliably without GPS. It was probably stupid to drop $7K on drone GPUs and all kind of gadgetry (6 battery bay for rapid charging, etc), but it was just really really hard to pilot around in Maryland (Montgomery County). I would constantly have it throw up warnings and alerts, even only hovering a few feet above the ground for small scale testing. I would have to disable the GPS to do small scale testing, and then with GPS enabled, it would straight up not allow me to pilot it. When I moved to Miami, brought it down there, but I managed to find an apartment right smack-dab in the MIA no-fly zone as well. The smaller drone was allowed to fly though, so I eventually got a small Mini 2 IIRC, which was a lot easier to pilot, but I was just so disappointed in not being able to use the larger scaled up version. I wanted to do realtime facial recognition (not at scale, just to show that commercial drones can be turned into research demonstrators) on the onboard GPU (apparently just a NVDA Jetson from 2017 era)

The irony is the M100 is genuinely great hardware - the payload capacity, the SDK access, the flight time with extra batteries. But DJI's geofencing treats the entire DC metro like a no-go zone, which makes sense from their liability perspective but means the thing is basically a $7K shelf ornament unless you want to deal with LAANC authorizations for every single flight.

I highly doubt this

Edit: owner of matrice m100 and a few other DJI drones

If you have access to mental health resources, I encourage you to leverage them.

They are turning it into a distributed system that you'll have to pay to access. Anyone can see this. CLI is easy to make and easy to support, but you have to invest in the underlying infrastructure to really have this pay off.

Especially if they want to get into enterprise VPCs and "build and manage organizational intelligence"

Sorry but that answer doesn't make any sense. Are you a bot?

Their system prompt + MCP is more of the culprit here. 16 tools, sophisticated parameters, you're looking at 24K tokens minimum

What is up with your strange claims on this page? https://michaelawhite.net/software/

There are so many contradictions in the "Claude Soul doc" which is distinct from this constitution, apparently.

I vice coded an analysis engine last month that compared the claims internally, and its totally "woo-woo as prompts" IMO

Sounds pretty much like today

Life, uh, finds a way

Feels like a breakup. Damn that sucks, can you get a mediator maybe?

Think of the children...

works fine for me, what's the matter?