Search Engineer
Open a video or resource link, watch it, then check its box to track your progress and unlock what comes next.
- Build the lexical search core, understand queries and measure relevance before adding ML
- Combine lexical and neural retrieval, then learn better ranking from judged evidence
- Operate search at scale and build a continuous relevance-improvement loop
Benchmark: Google Search / Search Ecosystems + Elastic Search Relevance / LTR + OpenSearch / Vespa retrieval engineering (Sep 2026) + CMU 11-642 Search Engines (Fall 2026) + Stanford CS276 Information Retrieval & Web Search Market refresh: search-engine systems + relevance - inverted indexes/BM25, query understanding, autosuggest, judgments, LTR/reranking, vector/hybrid retrieval, distributed indexing, latency/freshness, A/B loops and reliability. Complements Role 80 with deeper search implementation and relevance operations.
Search Live Openings
We don’t have a direct job-board feed matched to this one yet, so here are real, live US job searches for “Search Engineer” instead.
Build the lexical search core, understand queries and measure relevance before adding ML
Inverted indexes, analyzers + lexical ranking
Coverage: Information Retrieval / Text Analysis / BM25 / Indexing
Query processing, evaluation + autosuggest
Coverage: Query Processing / Relevance Metrics / Autocomplete / Long Tail
Combine lexical and neural retrieval, then learn better ranking from judged evidence
Hybrid lexical/vector retrieval + candidate generation
Coverage: BM25 / Embeddings / ANN-HNSW / Hybrid Search / RRF
Learning-to-rank, reranking + relevance judgments
Coverage: LTR / Feature Engineering / Reranking / NDCG / Click Signals
Operate search at scale and build a continuous relevance-improvement loop
Search infrastructure, vectors + long-tail reliability
Coverage: Distributed Search / HNSW / Quantization / Latency / Zero Results
Relevance experimentation + online improvement loops
Coverage: Judgment Sets / NDCG-MRR / UBI / A-B Testing / Observability
Role-Readiness Proof / Capstone
Complete a 5-day Search Engineering Lab on a 50k+ document/product corpus: define schema/analyzers; ship a BM25 baseline with filters, facets and autosuggest; create graded judgments; add ANN + hybrid retrieval and an LTR/reranker; measure Recall@K, MRR, NDCG@10 and zero-result rate; load-test latency/throughput/freshness; instrument query/click signals; diagnose 20 failing queries; ship one measured relevance gain with an A/B or interleaving plan, dashboards, SLOs and rollback.
Proof Rubric
- Working search system
- reproducible indexing
- sound query analysis
- measured lexical baseline
- judged relevance set
- correct offline metrics
- defensible hybrid/LTR gains
- no train/eval leakage
- latency, freshness and availability SLOs
- long-tail/zero-result failure analysis
- explainable ranking signals/business rules
- online experiment design
- rollback/runbook
- documented tradeoffs across relevance, latency, cost and maintainability