Founding Software Engineer

a16z-speedrun•San Francisco, CA

5d•Onsite

About The Position

Basis is building consumer-facing products that collect human interaction and expression data and transform user behavior into high-quality training signals for frontier AI labs. The company aims to help labs build more capable, grounded, and useful models while giving people a meaningful role in shaping those systems and rewarding them for their contributions. Basis says it works with leading labs, has a large contributor network, and collects recordings and preference votes at substantial scale. The Founding Software Engineer will build data pipelines and scalable systems for processing user recordings, managing rewards and advertising spend, and delivering curated data to clients. The role involves working directly with leading labs and collaborating closely with the founders in a fast-iteration environment, with high autonomy and end-to-end ownership. The company also describes related Research Engineer and MTS/Research-Software hybrid work, including training models, developing benchmarks, and publishing technical work. Basis is a small, fast-moving startup. The stated work hours are weekdays 10–8, and the company says it does not expect employees to work outside those hours. Roles and responsibilities are expected to evolve as the company grows. The position is based in a San Francisco office. Compensation includes equity, and the company offers 20 days of PTO, medical, dental, and vision insurance, meals on workdays, a company retreat, and immigration support.

Requirements

Enjoys system design, is intentional and detail-oriented, moves quickly with AI coding tools, loves building, and thrives on solving new problems.
For the related Research Engineer or hybrid scope: deep conceptual understanding of deep learning.
For the related Research Engineer or hybrid scope: experience fine-tuning strong open models.

Nice To Haves

For the related Research Engineer or hybrid scope: a track record of publications.

Responsibilities

Build pipelines that score, filter, and rank millions of user recordings each day using in-house machine-learning models and custom fraud-detection systems.
Design scalable systems from scratch to automatically control millions of dollars in user rewards and advertising spend.
Work directly with leading labs to assess their needs and deliver high-quality, heavily filtered recordings for frontier models.
Collaborate closely with founders in a fast-iteration environment and independently own systems end-to-end.
For the related Research Engineer or hybrid scope: train internal models, public benchmarks, and technical publications using multimodal data.
For the related Research Engineer or hybrid scope: collaborate with researchers at leading labs to create challenging open-source benchmarks for voice and interaction modeling.