Software Engineer, Agentic Search & Retrieval

Datasite is the essential infrastructure for M&A. Also comprising Grata, Blueflame AI, and Sherpany, we help deal breakers find opportunities, manage transactions, and make key decisions.

Whatever your talents, you can find a place for them here. Get involved in some of the biggest business ventures today. Work with other gifted people around the world. Follow your ambitions, or discover a whole new path.

Can you help us simplify the complex, create great experiences, and deliver better outcomes? Then let's talk.

Datasite's engineering organization builds the software the world's deal makers depend on to find, evaluate, negotiate, and close transactions: secure virtual data rooms, AI-assisted diligence, deal sourcing, and private-company intelligence.

We’re looking for hands-on Staff Software Enginee
r to own the context AI agents use on live deals. A wrong passage, a stale document, or a missed permission becomes a wrong answer in a high-stakes workflow. You will build the retrieval layer that makes agent answers grounded, current, and secure. You will work in Python and partner closely with Product and Data Science. You will stay hands-on on the decisions that change what an agent can know.

We are hiring in New York City and Minneapolis. Datasite is a hybrid company, with in-office collaboration on Mondays, Tuesdays, and Thursdays.

• Own retrieval for agents that work on live deals. Ground each answer in the right documents and data, and set the balance of quality, speed, and cost.
• Scale how the platform processes large documents and data tables, so an agent can still find and reason over content that does not fit in one pass.
• Shape the context agents depend on and evolve it as deal workflows get longer and more complex.
• Build the search tools agents call. Cover content that is indexed and content that is not indexed yet. Return the snippet, citation, and location a dealmaker can trust.
• Keep agent context aligned with the data room and with whoever may see each item. An agent must not read content the user cannot read.
• Decide when a model may answer from retrieved context and when the product must check the result.
• Mentor engineers and set technical direction for retrieval across teams.

• 6+ years of building and operating production software, with a record of leading complex technical work across teams.
• Strong hands-on Python, with experience in backend services, APIs, and large-scale data pipelines.
• Experience with production LLM applications that depend on retrieval: tool calling, RAG, embeddings, vector search, or context management.
• Strong foundations in permissions, multi-tenant data, and access control on what an agent can retrieve.
• Clear product and technical judgment. You can balance answer quality, latency, and cost, and explain the tradeoffs to engineering and product leaders.

• Long-running agent workflows that search or read documents.
• A vector store at production scale.
• Processing of large confidential documents and data tables.
• Hybrid search, citations, and evaluation of retrieval quality.
• Virtual data rooms, M&A documents, or similar document-heavy work.
• Tool interfaces that expose search to an agent.

The base salary range represents the estimated low and high end for this position based on a good faith assessment of the role and market data at the time of posting. Consistent with applicable law, each candidate’s compensation offer may vary and will be determined based on but not limited to, your geographic region, skills, qualifications, and experience along with the requirements of the position.

This position may be eligible for bonuses, commissions, or overtime if applicable. Benefits include…