SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
TypeSafe is an AI lab building production-grade AI systems that combine LLM reasoning with code reliability. The company recently launched Jev, their first System One Model, and is backed by a16z with $870M in Series A funding.
In this role, you will be a full-stack engineer responsible for building data-tooling interfaces, evaluation frameworks, and internal products that directly advance model intelligence. You'll work across the entire stack—Python backend, TypeScript/Next.js/Tailwind frontend, and Kubernetes orchestration—to create high-leverage datasets, user interfaces, and evaluation systems that unlock new capabilities.
Key responsibilities include:
- Design and build internal data-tooling and evaluation interfaces that make model development clearer (evaluation, debugging, data inspection)
- Develop and own evaluation frameworks to measure quality, reliability, and emergent characteristics across model iterations
- Create high-leverage datasets and products that unlock new use cases on top of the model
- Run rigorous analyses and experiments to understand how data, training choices, and interventions impact model behavior
- Transform raw model outputs and data into clear, inspectable UIs—owning domains end-to-end with craft and judgment
- Partner closely with product managers and customers to translate real-world needs into concrete model and system improvements
- Continuously iterate and co-discover new techniques for building trustworthy AI products
You'll be part of a small, flat, close-knit team working fully in-person from the San Francisco office near Embarcadero. The culture emphasizes emotional honesty, kindness, craftsmanship, and collaboration. The team values scrappiness, creative problem-solving in ambiguous spaces, and a healthy skepticism toward AI capabilities.
REQUIREMENTS:
- Solid full-stack engineering fundamentals with strong coding ability and product intuition
- Hands-on experience implementing LLMs; deep understanding of their capabilities and limitations
- Generalist mindset with ability to think beyond code to real-world system usage
- High attention to detail and strong quality bar
- Scrappy, hacky approach to problem-solving; comfort with ambiguity
- Skepticism toward LLM outputs; ability to critically evaluate model generations
- Ability to work fully in-person from San Francisco office
