TensorZero is an open-source platform that creates a feedback loop for optimizing LLM applications — turning production data into smarter, faster, and cheaper models.

1. Integrate our model gateway

2. Send metrics or feedback

3. Optimize prompts, models, and inference strategies

4. Watch your LLMs improve over time

It enables a data & learning flywheel for LLMs by unifying:

• Inference: one API for all LLMs, with <1ms P99 overhead

• Observability: inference & feedback → your database

• Optimization: from prompts to fine-tuning and RL (& even 🍓?)

• Experimentation: built-in A/B testing, routing, fallbacks

The project is brand new with an ambitious roadmap ahead. We started building in stealth in February, completed a successful technical pilot over the summer, and just announced the open-source project (September).

We’ve raised from FirstMark (backed ClickHouse), Bessemer (backed Anthropic), Bedrock (backed OpenAI), and many angels. We’re lucky to have years of runway, giving us the flexibility to fully focus on open source for now with an ambitious long-term vision.

Role

We are looking for a Founding Member of Technical Staff with a background in ML engineering or research.

Why not “engineer” or "researcher"? We’re not going to make a distinction between engineers and researchers to encourage people to have cross-functional scope and impact. Our CTO has a PhD in reinforcement learning but also writes infrastructure code in Rust.

Today we particularly need help building delightful interfaces, but expect everyone to contribute to the entire project over time. The vast majority of your work will be open source. You’ll have an opportunity to continue to master your current skills with the flexibility to learn new ones from scratch.

You can learn more about our technical roadmap and vision here. As a preview, if you joined today, you'd likely work on advanced inference strategies (e.g. MCTS) or new optimization recipes (e.g. different sorts of RL and APE).

Team & Culture

We’re a small technical team based in NYC (in person). As an early contributor, you’ll work closely with us and have a significant impact on the project’s future and vision.

• Viraj Mehta (CTO) recently completed his PhD from Carnegie Mellon, with an emphasis on reinforcement learning for LLMs and nuclear fusion, and previously worked in machine learning at KKR and a fintech startup; he holds a BS in math and an MS in computer science from Stanford.

• Gabriel Bianconi (CEO) was the chief product officer at Ondo Finance ($10B+ valuation) and previously spent years consulting on machine learning for companies ranging from early-stage tech startups to some of the largest financial firms; he holds BS and MS degrees in computer science from Stanford.

• Andrew Jesson (MTS) is an ML researcher with deep expertise in Bayesian ML, causal inference, RL, and LLMs. He’s finishing his post-doc at Columbia and previously received a PhD from Oxford, during which he interned at FAIR. He has 3.3k+ citations and several first-author papers at NeurIPS and other top ML venues.

• Aaron Hill (MTS) is a back-end engineer with deep expertise in Rust. He became one of ~30 people ever elected as maintainers of the Rust compiler… while still in college. Later, he worked on back-end infrastructure at AWS and Svix. He’s also an active contributor to many notable open-source Rust projects (e.g. Ruffle).

What We Offer

• Competitive compensation — We believe that great talent deserves great compensation (salary, equity, benefits), even at an early-stage startup.

• Open-source contributions — The vast majority of your work will be open source and public.

• Learning and growth opportunities — You’ll join with a background in ML but will have the opportunity (& be encouraged) to expand your skill set way beyond that (curious about Rust?).

• Small, technical, in-person team — You’ll work alongside a 100% technical team and help shape our vision, cul