Meridian Marketplace UK is hiring a Senior Senior Applied Scientist — Ranking & embeddings on the Search relevance team (Hybrid · London, United Kingdom). This is a hands-on role in search & recommendations—not a generic “AI enthusiast” post.
You will own ranking, retrieval, and online experimentation end to end: problem framing, offline evaluation, safe rollout, and iteration from production signals. Success is measured with clear metrics (quality, latency, cost, or business KPIs)—not slide decks alone.
Meridian Marketplace UK powers discovery for millions of buyers and sellers. Ranking blends classical signals with embeddings and fresh GenAI-assisted features—we need rigorous science and pragmatic experimentation.
Join our London hybrid office as a Senior Applied Scientist. You will collaborate with PMs, infra, and integrity teams to ship measurable relevance gains without compromising safety.
What you will do
- Design and analyse ranking / retrieval experiments: embeddings quality, calibration, exploration, and cannibalisation checks.
- Build offline metrics that correlate with marketplace KPIs; champion guardrails for cold-start and abusive patterns.
- Prototype new models (training and fine-tuning where justified) and coordinate production handoff with ML platform.
- Partner on A/B platform design: power analysis, ramp strategy, and automated monitoring for long-tail segments.
- Document methods and ship reusable datasets and feature libraries for the wider team.
What we look for
- MS/PhD in CS, stats, or related field—or equivalent industry track record in search / recommendations.
- Strong Python, SQL, and PyTorch (or similar); experience with large-scale feature stores and experimentation.
- Clear communication of causal and statistical nuance to product partners.
- 5+ years shipping ranking, search, or personalised feed systems at scale.
- Right to work in the UK.
Nice to have
- Exposure to two-sided marketplaces and fairness/safety constraints.
- Experience with Snowflake or BigQuery analytics loops.
- Publications or internal memos on embeddings or learning-to-rank.