You'll work on the parts of Duo that make it feel like a great salesperson: retrieval over store catalogs, tool use, evaluation and guardrails that keep it honest.
What you'll do
- Design prompts, tools and retrieval pipelines for production LLM features
- Build evaluation sets and measure quality, cost and latency
- Put guardrails in place so the assistant never invents facts
- Prototype new AI features with the product team
What we're looking for
- Experience shipping LLM-powered features to real users
- Strong Python or TypeScript
- Understanding of embeddings, retrieval and tool calling
- Rigour about evaluation — you measure before you claim
Nice to have
- E-commerce or conversion-rate experience
- Image generation or computer vision
- Experience with A/B or holdout testing