Member of Technical Staff, Research

The Token Company•San Francisco, CA

6d•Onsite

About The Position

The Token Company is a seed stage startup in San Francisco that has raised $12M from prominent investors. They train machine learning models that compress raw LLM inputs before they reach the expensive model, stripping out redundant parts to cut inference costs for enterprises building on LLMs. The team is small, with a tight research and product focus, and emphasizes shipping what they train.

Requirements

Trained models from scratch.
Owned data, architecture, and a training loop yourself, ideally in a research lab, a startup, or a scale-up.
Not just fine-tuning or calling an API.
Strong in ML fundamentals.
Fluent with transformers.
Learn fast and get to the core of a hard problem quickly.
Iterate on new ideas without hand-holding.
Care about production and seeing your model cut real inference costs for real users.

Nice To Haves

Pretrained a transformer, or done serious post-training or RL on one.
Built a novel architecture or training method with results to back it up.

Responsibilities

Own the model training stack end to end: data, architecture, training, evaluation, and shipping compression models into production.
Design and train models on NVIDIA B200s.
Turn a paper or a rough idea into a real training run.
Ship your trained model into something people actually use.