About the role
Every WAYO plan starts with a language model, then gets grounded in real place data and checked by a local. You'll own the part where AI meets reality: getting plans that respect opening hours, travel time, budgets and the things a traveller actually said.
Why now: Generic AI itineraries look great until you arrive at a place that closed last year. Our edge is plans that hold up on the ground, and that depends on how well we prompt, retrieve, verify and measure.
Your first 90 days
Day 30
Map how plans are generated today and build an evaluation set from real plans that locals corrected.
Day 60
Ship an improvement that measurably cuts the number of fixes locals have to make.
Day 90
Own the plan-quality metric and the roadmap for smart food and place suggestions.
What you'll do
- Design prompts, retrieval and tool calls that produce plans grounded in real places and opening hours.
- Build evaluation sets and automated checks so we know when a change makes plans better or worse.
- Use locals' corrections as training signal for better suggestions.
- Keep cost and latency per plan under control as usage grows.
- Work on SmartPulse, our live trip companion, so suggestions adapt to weather, crowds and time of day.
What you'll bring
- You've shipped at least one LLM-powered feature to real users.
- Strong Python or TypeScript.
- You measure model quality with evals instead of vibes.
- You can explain a trade-off between quality, cost and speed in plain words.
Bonus points
- Retrieval, ranking or recommendation systems experience.
- Experience with geo or place data, or with structured output from LLMs.
- You love travelling and have opinions about bad itineraries.
This probably isn't for you if…
- You only want to do research without shipping.
- You're not interested in measuring your own work.
Pay
Salary plus ESOPs. We share the exact range on the first call.