micro1 (https://www.micro1.ai) is an AI data company. It’s contracting ML engineers to build and check hard ML engineering tasks for an AI training project.
A typical task: implement or change a model, set up a reproducible training or inference run, cut memory use or raise throughput, debug a numerical or system-level failure. Then show the result meets objective correctness and performance targets.
What they ask for:
- An MS or PhD in CS, ML, AI, applied math, statistics, engineering, or a close field. Exceptional open-source or academic work can also qualify.
- Strong ML experience in industry or research
- Daily Python, plus comfort working with coding agents. The role is coding heavy.
- Tools like PyTorch, JAX, NumPy/SciPy, SGLang, vLLM, llama.cpp, Hugging Face Transformers and Tokenizers (equivalents are considered)
Terms:
- Listed rate: $80-$150/hour, paid per task that meets the spec (weekly minimum applies)
- About 15 hours a week. You pick the hours, weekends included.
- Fully remote. Check that your country is on the listing.
- Process: screening questions, a 30-minute AI interview, then a hiring manager review. micro1 says it often fills roles within 48 hours.
Apply (free, on micro1’s site):
Disclosure: this isn’t an official micro1 post, and I’m not micro1 staff. The link is my personal micro1 referral link, and I may receive a referral bonus if you’re placed. Applying is free and goes through micro1’s own site.
Alex Westfield / Micro1 Partner