Actively AI is focused on building superintelligent machines for Enterprise GTM organizations to enhance productivity. The Senior/Staff Software Engineer will design and scale the search and retrieval infrastructure that powers AI agents, ensuring data is accurately retrieved and utilized for optimal performance.
Responsibilities:
- Build the retrieval layer agents depend on. Design and scale the search and retrieval infrastructure that feeds Actively's agents, covering indexing, querying, ranking, and filtering across diverse customer data sources
- Turn raw, unstructured data into something retrievable. Design enrichment and entity extraction systems that pull structure, relationships, and context out of call transcripts, documents, and signals, making them queryable in ways that improve what agents actually see
- Own the Search for Agents Architecture: Define how data gets represented and stored, making deliberate choices about granularity, embedding models, and index configuration for different data types and use cases
- Build and iterate on ranking systems. Design and deploy reranking layers that maximize relevance for agent queries, and evolve them as data patterns and use cases change
- Develop shared retrieval primitives. Build the APIs and retrieval interfaces used by the Intelligence, Assistant, and Orchestration teams, balancing flexibility with consistency across consumers
- Own retrieval quality end to end. Build and maintain evaluation infrastructure using classical IR metrics, task-level success signals, and LLM-based techniques, catching regressions before they affect agent behavior
Requirements:
- Deep experience in search or retrieval systems. You have 5+ years building and operating retrieval systems in production, across multiple customers, data sources, or domains, and understand what relevance actually means at scale
- Background in information retrieval or applied ML. You've tuned relevance, deployed reranking strategies, and improved result quality in production, not just in experiments
- Understands the freshness problem. You've built retrieval pipelines over fast-changing data, including near-real-time indexing, incremental updates, or event-driven ingestion, and know how freshness trade-offs affect system design
- Comfortable with hybrid retrieval approaches. You've worked with systems that combine semantic search, keyword and lexical matching, and metadata filtering to balance recall, precision, and reliability
- Rigorous about evaluation. You've designed or evolved retrieval evaluation frameworks using IR metrics, task-level success signals, or automated quality checks, and you treat regressions as real incidents
- Thinks about retrieval architecture holistically. You know when to pre-compute versus retrieve at query time, how to manage index growth, and how to design retrieval paths that stay relevant as scale increases
- Prior experience at a search or retrieval-focused company (Elastic, Algolia, Cohere, Pinecone, Weaviate) or building shared search infrastructure used across multiple teams or products
- Experience with entity resolution, knowledge graph construction, or relationship extraction at scale, particularly over noisy or inconsistently structured source data