SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Exa is an applied AI lab building a next-generation search engine powered by massive-scale infrastructure, state-of-the-art embedding models, and high-performance vector databases. The company crawls the entire web continuously and powers search for Cursor, Cognition, HubSpot, and over 400,000 developers. Exa has raised $350M from Lightspeed, Benchmark, and a16z.
As a Research Engineer for Index Intelligence, you will own the signals and decisions that determine which pages are worth indexing, which links are worth following, which pages are duplicates, what has become stale, and what should never have been picked up. These indexing decisions have outsized impact on search quality—often more valuable than ranking work. The role involves designing and implementing learned signals that replace today's mix of heuristics and hand-picked thresholds.
You will work across crawling, indexing, and retrieval teams to ensure your signals drive real end-to-end improvements in search quality. Example projects include: improving parsing robustness on difficult pages and measuring the impact rigorously; building models to judge page quality and establishing shared definitions of quality across teams; modeling credibility and misinformation (source reliability, intent, freshness); and solving semantic deduplication to avoid collapsing pages users would want to see separately.
The role requires comfort with ambiguous problems where ground truth doesn't exist—your first job is defining what success means. You must think in terms of end-to-end impact: a signal only counts if it changes a decision and search improves measurably.
Desired Experience:
- Hands-on ML experience with classifiers, rankers, and calibration; strong data instincts at web scale
- Comfortable owning problems with no ground truth and defining what the right answer means
- End-to-end impact mindset: signals must drive decisions that improve search
- Cross-functional collaboration across crawling, indexing, and retrieval
- Deep interest in the problem of finding high-quality knowledge and its importance for the world
