Home
micro1

Member of Technical Staff, Model Training (post-training research)

  • Pay: $7 to $8 / hour
  • Contract: Full time
  • Who can apply: Open worldwide
  • Posted Oct 27, 2026

In short

Good news: this is a full-time remote job paying 7 to 8 USD per hour. You would dig into post-training research on large language and multimodal models, testing new ways to make AI reason better. If you love experiments and finding out what actually works, this one is made for you.

What you will do

  • Run post-training experiments on large language and multimodal models, using methods like supervised fine-tuning, reinforcement learning and distillation.
  • Tweak training recipes by testing different data mixes, sampling methods and evaluation setups.
  • Learn from expert demos, human feedback and corrections to spot where people help the model the most.
  • Design controlled tests and ablation studies, then double-check your results.
  • Take a research idea from first sketch all the way to validated results.

What you need to apply

  • Solid grounding in machine learning, deep learning and modern training techniques.
  • Hands-on experience fine-tuning or post-training large-scale language or multimodal models.
  • Strong Python and PyTorch skills to build, debug and speed up training pipelines.
  • A rigorous, reproducible approach that focuses on measurable improvements.

Who can apply

Open worldwide, since the posting states no country restriction. You can apply from anywhere as long as you match the skills above.

A tip from Romowork

We suggest showing one concrete experiment you ran, with clear results, in your application. This team values measurable improvements, so numbers and proof will help you stand out.

Originally published on micro1

Related jobs