Gather is building the all-in-one work communication tool - think of it as a Zoom and Slack replacement, built for an AI-native world. Grapevine is Gather’s newest product, which provides rich context for AI agents to navigate all the tools used for work (think: Slack, Notion, GitHub, etc.) and structure the knowledge for agents. This is the holy grail opportunity for the AI ecosystem, and we believe we have unique insight to pull it off. As a Staff Machine Learning Engineer on Grapevine, you'll be working on building out the core search algorithm and evals for our knowledge base system. There are novel, as-yet-unsolved technical challenges here in building search algorithms for agents, rather than humans.
Requirements
- Extensive experience with search & retrieval systems (esp. Opensearch, vector stores, PSQL)
- Experience on algorithmic engineering problems
- Experience building production-grade RAG systems (knowledge graphs, etc.)
- Comfort with hypothesis-driven research environments
- Intellectual curiosity that drives you to explore new approaches and stay current with emerging technologies
- Strong debugging and investigation skills; you’re comfortable digging into an ambiguous problem statement and coming up with solutions
- A keen intuition for well-designed software and well-written code
Responsibilities
- Design and implement the core search algorithm across unstructured knowledge, improving the quality of Grapevine’s ability to reason over difficult questions
- Design increasingly complex and general evaluations to define new goals to surpass
- Tune how new information sources are incorporated into Grapevine’s knowledge
- Create “digestion” systems to extract vital information from unstructured data
- Mentor and inspire other engineers to think creatively about AI applications
Other
- Experience building internal tools for cross-functional organizations
- Excitement about working on a small team where your contributions have an outsized impact
- Intrinsic motivation and the drive to own and solve difficult problems without guidance
- Excellent interpersonal and communication skills, with technical leadership abilities
- A healthy tolerance for the chaotic balance between Doing Things Right and iterating and shipping quickly
