Product Manager, RAG Architecture

About the role

Most businesses we work with want AI that knows their documents, data and processes, and answers with sources they can check. Retrieval augmented generation is how we get there, and doing it well is a product problem as much as an engineering one: what to index, how to keep it fresh, who may see what, and how to know an answer is right.

As the Product Manager for RAG Architecture, you will define how NerfedLab builds knowledge grounded AI across our products and client work. You will work closely with AI engineers and client teams, and turn what works into a platform we can deliver again and again.

What you will do

• Own the vision, roadmap and delivery of our retrieval and knowledge products, from ingestion and indexing to search, generation and citations
• Define quality with engineering: build evaluation sets, choose measures for relevance, faithfulness and coverage, and hold the bar
• Make architecture trade offs with engineers across chunking, embeddings, hybrid search, re-ranking, caching and cost
• Work with enterprise clients to understand their data, permissions and compliance needs, and turn them into requirements
• Design the access control, freshness and audit features that make AI answers safe to rely on inside a company
• Follow the research and the market, and keep an honest view of where our approach leads and where it lags

You may be a good fit if you

• Have 5+ years of product management experience, with at least 2 years on AI, search or data products
• Understand RAG systems in depth, including embeddings, vector and keyword search, re-ranking and evaluation
• Have shipped a search, knowledge or AI product to enterprise customers
• Are technical enough to read code, write SQL and Python, and debate an architecture with engineers
• Can turn vague needs from many stakeholders into a clear plan, and say no with reasons
• Communicate clearly in writing and in front of clients

Strong candidates may also have experience with

• Vector and search engines such as pgvector, Pinecone, Weaviate or Elasticsearch
• Knowledge graphs, document parsing or agentic retrieval
• Regulated industries such as finance, health or government
• A background in engineering or data science

About NerfedLab

NerfedLab builds AI products, helps businesses put AI to work in their everyday operations, and runs a research team that publishes its work in the open. Zocri, our product, lets anyone make an app or a website by describing it, then run the business behind it: sign-in, payments, messages and marketing in one place.