Software Engineer - AI Research

The Role

Design and deliver the intelligence layer that turns enterprise process knowledge, documents, interviews, system logs, into a governed, provenance-carrying process ontology, and the AI systems that extract, consolidate, ground, and continuously learn from it. Ensure every output is evidence-backed; hallucination is a product failure, not a metric.

What You’ll Do

• Build end-to-end pipelines for ingesting complex documents, handling layouts, tables, and scanned diagrams, and generate robust embeddings for long-form content

• Design schema-constrained LLM extraction frameworks using typed schemas, validation loops, and automated repair/re-prompt cycles

• Own retrieval engineering solutions, integrating hybrid search, reranking, and query expansion to surface relevant context efficiently

• Deliver knowledge-graph construction and entity resolution pipelines, modeling property-graphs in Neo4j or equivalent stores and enabling GraphRAG retrieval

• Improve grounding mechanics by integrating terminology stores and retrieval-in-the-loop generation to ensure evidence-backed outputs

• Collaborate with cross-functional product and research partners to define evaluation datasets, grading rubrics, and regression gates, feeding field insights back into the roadmap

• Build multi-agent orchestration using planner/worker patterns and tool routing on native SDKs, ensuring seamless state sharing and coordination

• Own MCP and agent-to-agent integration, implementing secure contracts, authentication boundaries, and context passing between services

What You’ll Bring

• Proven expertise delivering production-grade AI solutions in Python, with strong typing, test coverage, and async pipeline design

• AI-native velocity as a default mode of working

• Deep knowledge of large-language-model prompting, schema-driven extraction, and automated repair loops

• Mastery of retrieval techniques, including dense embeddings, hybrid search, reranking, and query expansion

• Experience building and operating property-graph databases such as Neo4j, and implementing entity resolution and GraphRAG pipelines

• Hands-on competence with grounding architectures, terminology stores, and retrieval-in-the-loop generation

• Proficiency in multi-agent orchestration frameworks (e.g., LangGraph, OpenAI Agents SDK) and secure MCP or A2A integration patterns

• Track record of defining evaluation datasets, constructing graders, and establishing regression testing with tracing tools like Langfuse

• Strong judgment to balance research ambition with product constraints, and clear communication with engineers, product managers, and senior stakeholders

• Master's degree in Computer Science, AI, or a related field (or equivalent practical experience), with 2+ years of relevant experience.