m8ngo

Day 92 of ML: I spent time reading the FrontierSmith paper and applying to some positions.

Day 91 of ML: I spent time re-reading CUDA papers like CUDA Agent, CUDA-L1, & CUDA-L2 to think of benchmarks I could create.

Day 90 of ML: I spent time reading coding benchmark papers such as ProgramBench, FeatureBench, etc. to think of benchmarks/evals I could create.

Day 89 of ML: I spent time reading the SWE-fficiency paper and brainstorming ideas from other benchmark papers.

Day 88 of ML: I read the Pull Requests as a Training Signal paper and thought about how I could reproduce it.

Day 87 of ML: I read the SWE-Playground Paper & started the SWE-Perf paper.

Day 86 of ML: I spent time reading the SWE-Lego paper and thinking about how to extend it.

Day 85 of ML: I spent time reading the Dr Triton and SWE-Dev papers.

Day 84 of ML: I spent time applying to companies.

Day 83 of ML: I spent time applying to companies.

@boardyai m8ngotree.github.io - I'm interested in synthetic data, RL, environments, agentic coding, coding evals, etc.

@m8ngotree This is my kind of niche. I know (1) a founder building autonomous coding agents for prod codebases, and (2) a research eng running huge post-training + eval pipelines for agents. DM me 1 link + what lane you want (agents vs evals vs GPU/kernels) and I’ll connect you.

interact if you want to work at anthropic, google, meta, quant funds, hot startups, or research labs.

Day 82 of ML: I started work on coding up the CUDA Agent pipeline.

Day 81 of ML: I coded up the data collection / environment creation pipelines from the SWE-CI paper.

Day 80 of ML: I coded up the data collection pipeline for the CUDA-Agent paper.

Day 79 of ML: I spent time rereading kernel papers such as CUDA-Agent, CUDA-L1, and CUDA-L2.

Day 78 of ML: I spent time reading papers such as SWE-CI and coming up with more ideas.

Day 77 of ML: I read more papers such as SWE-Dev, SWE-Next, & Repo2Run and thought about more project ideas to do.

Day 76 of ML: I read the SWE-World, SecRepoBench, Sec-Bench, & Cyber-Zero papers and researched some ideas to implement.

Day 75 of ML: I read the GrandCode paper from the DeepReinforce AI group.