$200,000 - $300,000
San Francisco, Hybrid
Permanent / Full-time
A product-led AI start-up is building one of the most widely adopted AI work companions in the world, operating at massive real-user scale with millions of professionals relying on it daily. The challenge problem now is designing AI systems that reliably support complex knowledge work across preparation, collaboration, and follow-through, inside products people trust. This role is ideal for someone who wants to work across AI engineering, product thinking, and ultimately shape how AI actually shows up in day-to-day professional workflows.
Why This Role Matters
- Own how AI supports high-stakes knowledge work
- Design multi-step AI workflows that users rely on repeatedly
- Help define how agent-like systems behave inside a consumer-grade product
- Work beyond prompt design into evaluation, iteration, and reliability
What You’ll Do
- Own the end-to-end design of AI-first workflows for preparation, collaboration, and follow-up
- Design and iterate multi-step LLM / agentic systems, spanning intent understanding, planning, tool invocation, memory usage, and refinement loops
- Build reusable AI skills, prompts, templates, and evaluation pipelines that can power multiple product experiences
- Define success metrics for AI behaviour, run experiments, and use real interaction data to improve usefulness and reliability
- Partner closely with engineering and ML teams to ship quickly while maintaining a high bar for product quality and user experience
What You Bring
- Proven experience shipping AI/ML powered products end to end
- Strong working understanding of LLM systems: prompting, tool calling, retrieval, context construction, evaluation, and common failure modes
- Ability to translate user needs into clear flows, specs, and examples, including edge cases and expected behaviours
- Comfort working directly with data and interaction logs to debug issues and compare variants
- Hands-on experience designing agent-like workflows involving multi-step plans, multiple tools, and refinement or self-correction
