About TypeSafe
TypeSafe AI is an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software by combining the intelligence of LLMs with the efficiency and reliability of code into a new shape of AI: System One Models. Based in San Francisco, TypeSafe AI recently launched its first public model, Jev.
While others chase benchmarks and academic puzzles, we’ve been quietly rethinking the LLM stack from first principles — building a new kind of general frontier model designed for real-world reliability, decision-making, and autonomy in production.
We’re a small, fast-moving team from OpenAI, Google Brain, and Meta/FAIR, backed by top-tier investors. Since mid-2024, we’ve been engineering the foundation for what comes after the current “state-of-the-art” — a model that actually gets things done.
About the role
We're looking for scrappy engineers who ship data-tooling interfaces fast. You will progress model intelligence through the interfaces, datasets, and evaluation tooling you build, combining product sense, data science, and engineering. You'll be part of the real "secret sauce" of TypeSafe: figuring out how our models can provide production-ready reliability in the real world.
Our tech stack is primarily Python. We also use TypeScript, Next.js, and Tailwind CSS for frontend, with Kubernetes for orchestration. We empower developers to use any tooling they find helpful for getting their job done, including Claude Code and Cursor.
What you'll do
Create high-leverage datasets, products, and user interfaces that unlock new capabilities and use cases on top of our model
Build internal data-tooling and evaluation interfaces — fast — that make model development clearer (evaluation, debugging, data inspection)
Develop and own evaluation frameworks to measure quality, reliability, and emergent characteristics across model iterations
Run rigorous analyses and experiments to understand how data, training choices, and targeted interventions impact model behavior
Turn raw model outputs and data into clear, inspectable UIs — owning a domain end-to-end, with craft and judgment over simple optimization
Partner closely with product managers and customers to translate real-world needs into concrete model and system improvements
Continuously iterate, learn, and co-discover new techniques for building exceptional, trustworthy AI products
We are looking for people who
Are generalists (~2–7 yrs) with solid fundamentals, strong coding ability, and product intuition
Maintain high attention to detail and a strong bar for quality
Enjoy thinking beyond the code to how systems are used in the real world
Are scrappy and hacky, and thrive in ambiguous problem spaces — taking satisfaction in finding creative solutions
Don't trust LLMs blindly — they look at the reality of model generations
Have hands-on experience implementing LLMs and understand their capabilities and limitations
Life at TypeSafe
We’re a small, flat, close-knit team working to make intelligence dependable enough to become part of everyday software. We work fully in person from our San Francisco office near Embarcadero station. We love what we do and care deeply about the work.
We strive for excellence and craftsmanship and won’t stop until we get there. When the team wins, we all win, and we enjoy collaborating and inspiring each other to grow—as a team and as individuals.
We value emotional honesty, kindness, and bringing your whole self to work. We build machines; we don’t try to be machines.
We want TypeSafe to be the place where you do the most impactful work of your career and help define our future as a company.
We provide
