Staff Retrieval Engineer

Relativity

20h•Hybrid

About The Position

We are seeking a Staff Retrieval Engineer to join the Retrieval Engineering group at Relativity. This role is ideal for a deeply technical leader in information retrieval who thrives on designing large-scale search systems, optimizing retrieval infrastructure, and advancing search quality and performance across our platform. As a Staff Engineer, you will play a key role in defining and evolving our retrieval architecture, shaping how we index, store, and surface data across billions of legal documents. You will build next-generation search capabilities that blend traditional IR with modern vector search and AI-driven approaches. Your impact will span multiple teams, and you’ll collaborate with architecture, product, and data science leaders to ensure our retrieval stack is scalable, resilient, and aligned with both developer and customer needs. This is a high-impact role for someone who combines expert retrieval engineering capabilities with strategic thinking, mentorship, and a passion for building the future of intelligent search in a cloud-native environment.

Requirements

8+ years of professional experience in software engineering, with significant focus on information retrieval systems at scale.
Deep expertise in search engines and frameworks (Elasticsearch, Solr, Lucene, Vespa, OpenSearch, or equivalent).
Strong knowledge of retrieval models (BM25, vector similarity, hybrid retrieval, learning-to-rank, neural reranking).
Proven experience with distributed systems and storage, including index sharding, replication, and consistency trade-offs.
Strong programming skills in Java, C++, C#, Python, or Go and experience with performance optimization at the system level.
Proficiency with data processing frameworks (Spark, Flink, Kafka, Kinesis) for indexing and retrieval pipelines.
Track key retrieval metrics such as accuracy, latency, and fallback rate.
Experience operating retrieval systems in cloud-native environments (Azure, AWS, or GCP), including containerization (Docker, Kubernetes) and CI/CD.

Nice To Haves

Experience integrating vector databases (Pinecone, Weaviate, Milvus, FAISS, or pgvector) into production retrieval systems.
Familiarity with large-scale machine learning for ranking: embeddings, transformers, reinforcement learning from user feedback.
Understanding of privacy, compliance, and security requirements in enterprise search.
Experience with observability stacks (Prometheus, OpenTelemetry) applied to retrieval systems.
Experience with knowledge graph technologies (Neo4j, JanusGraph, TigerGraph, RDF/SPARQL, GraphQL, or property graphs) and their integration into hybrid retrieval systems.
Familiarity with legal tech, e-discovery, or enterprise SaaS search challenges.

Responsibilities

Architect, design, and optimize retrieval infrastructure at scale, including indexing pipelines, query execution frameworks, and storage layers.
Lead the evolution from traditional inverted-index search to hybrid retrieval systems that combine symbolic (BM25, learning-to-rank) and semantic (vector search, embeddings, RAG) approaches.
Drive adoption of retrieval best practices: query understanding, ranking models, caching, index sharding, distributed execution, and relevance evaluation.
Build fault-tolerant ingestion and indexing pipelines leveraging event-driven and microbatch architectures.
Collaborate with AI/ML engineers to integrate LLM-augmented retrieval, query expansion, re-ranking, and feedback loops into production search flows.
Partner with platform teams to ensure retrieval systems are observable, performant, and cost-efficient across multi-tenant Kubernetes clusters.
Establish benchmarking and evaluation frameworks for precision, recall, latency, and query coverage, and drive continuous improvement in retrieval quality.
Contribute to strategic technical decisions that shape Relativity’s future search capabilities and ensure they scale with the growth of our data and customers.
Incorporate knowledge graph–driven retrieval by modeling legal entities and relationships, integrating graph queries with text/vector search, and applying KG features to improve ranking and explainability
Mentor engineers across teams, lead design reviews, and champion technical excellence in search and retrieval.