Sign in

Matthew Kenney

@baykenney.bsky.social
1.3K followers 1.2K following 36 posts

Founder - Algorithmic Research Group. Previously: Senior ML Engineer at Apple, Asst. Research Prof at Duke University, Duke Data Science | PSU '15 | Cornell '11

PostsRepliesMedia
Reposted by Matthew Kenney
algoresearch.bsky.social @algoresearch.bsky.social · 02/04/2026
There's rightly a lot of excitement around Karpathy's autoresearch. We've been studying at ARG for a couple years now: what happens when you put an AI agent in a loop and let it run experiments, evaluate results, and iterate without you. We've built a bunch of benchmarks and tools to measure this. 🧵
101
Matthew Kenney @baykenney.bsky.social · 02/04/2026
bsky.app/profile/algo...
000
Matthew Kenney @baykenney.bsky.social · 02/04/2026
Over the past ~2 years I’ve been working hard on agents, models, and datasets to understand what recursive self-improvement might look like, and what path supporting infrastructure for this line of research might take. Very excited to open source some of that work starting today
130
Matthew Kenney @baykenney.bsky.social · 29/07/2025
Very excited to launch this little tool that we’ve been building. ScoutML is an API built for AI researchers and agents that includes a ton of metadata on each paper. It’s been super helpful for us as we run our research agents internally. x.com/algoresearch...
x.com
Algorithmic Research Group on X: "At ARG, we're laser-focused on understanding recursive self-improvement. We're confident that as models scale, RSI will accelerate the frontier of AI at ever-increasing speeds. Over the past year, we've created benchmarks, agents, and AI systems to measure how this might happen. https://t.co/JyOPFSB8DJ" / X
At ARG, we're laser-focused on understanding recursive self-improvement. We're confident that as models scale, RSI will accelerate the frontier of AI at ever-increasing speeds. Over the past year, we've created benchmarks, agents, and AI systems to measure how this might happen. https://t.co/JyOPFSB8DJ
030
Matthew Kenney @baykenney.bsky.social · 19/05/2025
If you're working on ML and this resonates, I’d love to hear what you'd want it to do. We're opening up a limited beta. Link below: prospectml.com
prospectml.com
ProspectML | A research assistant that helps researchers generate insights and code to accelerate their work.
A research assistant that helps researchers generate insights and code to accelerate their work.
011
Matthew Kenney @baykenney.bsky.social · 19/05/2025
It’s built on top of a foundation of parsed metadata from papers, code, and repos—models, metrics, datasets, SOTA claims, GPU counts (and types), ablation studies, citations, etc. It’s already become crucial to our internal research, and we hope it can be helpful to others, too.
100
Matthew Kenney @baykenney.bsky.social · 19/05/2025
It’s designed to support that murky, nonlinear part of the research process, where you're still figuring out what's interesting.
100
Matthew Kenney @baykenney.bsky.social · 19/05/2025
You give it a question like “How can we improve generalization in low-resource RL?” and it returns distilled insights, speculative ideas, and experimental code. Not final answers, just something to push the thinking forward.
100
Matthew Kenney @baykenney.bsky.social · 19/05/2025
Most of the time, I end up manually digging through papers, chasing links, and piecing together ideas. It works, but it’s slow, and it doesn’t scale with curiosity. I’ve been trying to fix that with a platform we're building called ProspectML.
100
Matthew Kenney @baykenney.bsky.social · 19/05/2025
A lot of ML tools help you implement. Not many help you think. When I’m exploring a new research direction, I don’t want another search engine or citation graph. I want something that’s actually read the literature, can suggest promising directions, and helps me reason through tradeoffs.
141
Reposted by Matthew Kenney
algoresearch.bsky.social @algoresearch.bsky.social · 11/01/2025
hello world!
021
Matthew Kenney @baykenney.bsky.social · 11/01/2025
ARG is on Bluesky! Please follow here: @algoresearch.bsky.social
000
Matthew Kenney @baykenney.bsky.social · 11/01/2025
good post on 2025 ai safety research directions: alignment.anthropic.com/2025/recomme...
alignment.anthropic.com
Recommendations for Technical AI Safety Research Directions
011
Matthew Kenney @baykenney.bsky.social · 26/12/2024
Back in Pennsylvania, drinking schuylkill county coal cracker (boilo) and making pierogies
030
Matthew Kenney @baykenney.bsky.social · 30/11/2024
That’s because it was from 2022
110
Reposted by Matthew Kenney
Juan Mateos-Garcia @jmateosgarcia.bsky.social · 26/11/2024
AI for science could be more impactful than chatbots. It is already helping win Nobel prizes and accelerating drug development and materials discovery. Today we published an essay about it: why it matters, how it’s happening and its implications. Here is a summary from an econ / social sci lens.
27930
Reposted by Matthew Kenney
Chris Paxton @cpaxton.bsky.social · 28/11/2024
Important point that the open protocol makes extracting data from bluesky easy. Can't have it both ways. I like the protocol and think this site is well designed, but that means anyone can and will analyze these posts (if there is value to them, which I'm honestly less convinced of than some)
0112
Matthew Kenney @baykenney.bsky.social · 28/11/2024
071
Reposted by Matthew Kenney
Stella Biderman @stellaathena.bsky.social · 28/11/2024
A dataset of 1 million or 2 million Bluesky posts is completely irrelevant to training large language models. The primary usecase for the datasets that people are losing their shit over isn't ChatGPT, it's social science research and developing systems that improve Bluesky.
825139
Matthew Kenney @baykenney.bsky.social · 28/11/2024
Wait what even is this platform. This is insane
130
Matthew Kenney @baykenney.bsky.social · 25/11/2024
What! If it works for umap-learn vs umap i’m in.
000
Matthew Kenney @baykenney.bsky.social · 25/11/2024
huggingface.co/AlgorithmicR...
huggingface.co
AlgorithmicResearchGroup (Algorithmic Research Group)
Org profile for Algorithmic Research Group on Hugging Face, the AI community building the future.
000
Matthew Kenney @baykenney.bsky.social · 25/11/2024
I have a community project in Eleuther and open source all of my research: bsky.app/profile/bayk...
110
Matthew Kenney @baykenney.bsky.social · 25/11/2024
Jk the rest are great. Just a big uncle nearest fan
110
Matthew Kenney @baykenney.bsky.social · 25/11/2024
5 . . . . . . . . 3 2 4 1
110
Matthew Kenney @baykenney.bsky.social · 24/11/2024
We welcome PRs, contributions, additional tasks, and task revisions. Excited to see how agents perform on this benchmark.
000
Matthew Kenney @baykenney.bsky.social · 24/11/2024
We develop a baseline agent, with tools for coding, research (via Semantic Scholar), and model training, built on top of Sonnet 3.5 and GPT-4o. Our baseline agent performs well across tasks, but generally fails to move beyond baseline implementations.
100
Matthew Kenney @baykenney.bsky.social · 24/11/2024
ML Research Bench adapts tasks from ML conference competitions like ‘NeurIPS Large Language Model Efficiency Challenge: 1 LLM + 1GPU + 1Day’ and ‘LLM Merging Competition’. We prompt agents to complete these challenging tasks. These tasks move beyond simple ML tasks.
100
Matthew Kenney @baykenney.bsky.social · 24/11/2024
(re-posting from X) Can we get AI to accelerate AI research and development? I’m excited to release ML Research Benchmark, an agentic benchmark of 7 ML conference competition tasks. Paper: arxiv.org/abs/2410.22553 Tasks: github.com/AlgorithmicR... Agent: github.com/AlgorithmicR...
arxiv.org
ML Research Benchmark
Artificial intelligence agents are increasingly capable of performing complex tasks across various domains. As these agents advance, there is a growing need to accurately measure and benchmark their c...
220
Matthew Kenney @baykenney.bsky.social · 23/11/2024
Maxo Kream
010
Matthew Kenney @baykenney.bsky.social · 23/11/2024
Sure! Multi-role or multi-module. From a software development perspective I think of them as microservices. Nothing more
010
Matthew Kenney @baykenney.bsky.social · 23/11/2024
There’s a paper, I’ll try to find it
100
Matthew Kenney @baykenney.bsky.social · 23/11/2024
Idk about ‘strengths’ and ‘perspectives’ but you do want a separation of concerns if your agents have a lot of tools. And you do want specific system prompts to guide their objectives. If you pack too many tools into an api call, the model will only use a handful of them.
100
Matthew Kenney @baykenney.bsky.social · 23/11/2024
For example a multi-agent system that conducts ai research, you have an agent for research, one for writing and running code, one for monitoring training runs, one for evaluating results. You can craft the system prompts and tools for each agent wrt their role.
010
Matthew Kenney @baykenney.bsky.social · 23/11/2024
If you have a single agent that uses tools (function calling), and you give them access to many specific tools, they tend to only use a few of them. By using a multi-agent system you have a separation of concerns where the agent only has to manage a few tools.
100
Reposted by Matthew Kenney
Michael Tran @huytransformer1.bsky.social · 22/11/2024
AI Conferences @neuripsconf.bsky.social @cvprconference.bsky.social @iccv.bsky.social @eccv.bsky.social @iclr-conf.bsky.social @blog.neurips.cc.web.brid.gy go.bsky.app/45EuhSi
43714
Matthew Kenney @baykenney.bsky.social · 22/11/2024
Looking forward to checking it out! I don’t know either but have been wanting to pick up a functional language
110
Matthew Kenney @baykenney.bsky.social · 22/11/2024
I somehow ended up following a ton of OCaml people and now I’m like I should probably learn OCaml
2191
Matthew Kenney @baykenney.bsky.social · 22/11/2024
Add me please! 🙋🏾‍♂️
110
Reposted by Matthew Kenney
Dan Roy @roydanroy.bsky.social · 18/11/2024
I've created an initial Grumpy Machine Learners starter park. If you think you're grumpy and you "do machine learning", nominate yourself. If you're on the list, but don't think you are grumpy, then take a look in the mirror. go.bsky.app/6ddpivr
12441355
Matthew Kenney @baykenney.bsky.social · 15/11/2024
ESP32 with wifi actually
050
Matthew Kenney @baykenney.bsky.social · 13/11/2024
Yes please!
120
Matthew Kenney @baykenney.bsky.social · 12/11/2024
Hello world
290