R&D Engineer

Overview

The R&D Engineer is a researcher-practitioner - someone equally comfortable reading papers and shipping prototypes. They are responsible for staying close to the frontier of LLM and AI systems research, identifying ideas with practical value, and building POCs that demonstrate feasibility. These POCs are demoed to the team and, where validated, handed off to the Platform and AI Engineers to productionize. This person does not work in isolation - they collaborate closely with the rest of the team throughout the research-to-product pipeline.

What You'll Do:

• Monitor and synthesize developments in LLM systems, AI infrastructure, and adjacent research areas
• Prototype and evaluate novel ideas with a fast feedback loop - build, demo, iterate
• Extend and improve the existing agent and RAG SDKs with new retrieval strategies, agent patterns, and provider integrations
• Evaluate new embedding models, reranking approaches, and retrieval architectures against the existing pipeline
• Collaborate with the Platform and AI Engineers to hand off validated POCs
• Produce clear documentation and demos that make research accessible to non-researchers
• Contribute to discussions with the tech lead on long-term research direction and where to place technical bets
• Engage with the broader research community - conferences, papers, open-source projects
• Identify opportunities where emerging techniques can address real problems surfaced by the AI Engineer

What You Bring:

• Hands-on experience building or evaluating LLM systems - fine-tuning, inference, retrieval, or agents
• Familiarity with RAG architectures beyond naive vector search - HyDE, MMR, graph-based retrieval, reranking, or hybrid search
• Exposure to agent frameworks and multi-agent orchestration patterns (PydanticAI, LangGraph, AutoGen, CrewAI, or similar)
• Demonstrated ability to move quickly from idea to working prototype
• A portfolio of explored ideas - published work, technical blog posts, side projects, or OSS contributions
• Strong communication skills - can explain complex research to non-researchers clearly
• Collaborative by default; understands that research value is realized when ideas ship
• Core: Python, agent frameworks (PydanticAI or equivalent), RAG pipeline design (embedding, retrieval, reranking), managed ML platforms (Vertex AI or equivalent)
• Nice to have: Systems-level programming experience, familiarity with inference infrastructure (vLLM, TensorRT), experience with RLHF or fine-tuning pipelines, MCP protocol, graph databases (Neo4j)