FingerTrip · Founding team
AI Recommender Systems Engineer
Build the intelligence that makes FingerTrip personal. Develop generative recommender systems that combine long-term taste with real-time context — location, time, weather, budget, mood, and company — to help people decide where to go next.
How we work
We work primarily on Mac, with Notion as our shared workspace and Claude, Codex, or Cursor as everyday building tools. Experience with them is welcome; curiosity and a willingness to learn them are essential.
What you will do
- Develop hybrid recommendation systems spanning retrieval, ranking, embeddings, and generation.
- Learn from visits, saves, skips, dislikes, explicit feedback, and context, while grounding every result in real places and real constraints.
- Build evaluation, experimentation, feedback, diversity, and safety loops alongside production inference and data pipelines on GCP or AWS.
- Turn model quality into product quality, deciding with design and engineering how a recommendation is explained and when the agent should stay quiet.
What we are looking for
- Strong Python and applied machine-learning experience with recommender systems, ranking, retrieval, embeddings, or personalization.
- Experience shipping LLM or generative systems and measuring them with offline metrics and online experiments.
- A practical understanding of cold start, bias, diversity, multi-objective optimization, latency, reliability, and cost.
- The judgment to recognize when a simpler model, a better feature, or cleaner data beats a larger one.
Nice to have
- Experience with GCP or AWS ML workloads, PyTorch, JAX, TensorFlow, vector search, feature pipelines, or model serving.
- Experience in geospatial or local search, travel, contextual recommendations, human feedback, reinforcement learning, or agentic systems.
- Published work, open-source contributions, or competition results in recommendation or ranking.
Why FingerTrip
Search returns results, maps return places, social returns inspiration, and travel apps return inventory — and people are still left to turn all of it into a decision themselves. We believe AI changes that interface for good. FingerTrip understands what someone actually wants and helps them go and do it, and every founding role here shapes what that becomes.