Junior AI Engineer | Financial Services | Venture Capital

About this role

ABOUT BIP CAPITAL BIP Capital is an integrated private market investment platform built to create and capture opportunities through BIP Ventures traditional venture anchor funds, an Evergreen equity BDC, and private credit offerings. With a distinctive multi-stage, multi-sector investment approach and a growing array of capital offerings, BIP Capital has generated consistent top quartile returns since 2009. OUR PHILOSOPHY We are champions of our investors’ goals and stewards of founders’ dreams, but more than that, we are ethical, honest partners who serve, educate, and protect our founders and investors. We take on the challenges and complex conversations because we operate out of integrity and genuine care. ABOUT THE ROLE BIP Capital's technology team is hiring a Junior AI Engineer to help bring large language models into the day to day work of an investment platform. You'll work directly with a Senior AI Engineer to add AI capabilities to our existing full-stack applications, and to design and ship retrieval-augmented generation (RAG) systems, agentic workflows, and prompt-engineered LLM pipelines that real investment and operations teams rely on. This is an apprenticeship-style role built for someone early in their career who is a strong full-stack engineer first and genuinely curious about applied AI. You will not be expected to own architecture on day one. You will be expected to execute well, learn quickly, and take real work off the Senior AI Engineer's plate within your first few months Office Environment: Hybrid / On-site (In-office Wednesdays). What you'll do Add AI capabilities to our existing full-stack applications: new LLM-powered features, workflows, and interfaces built directly into the products our investment and operations teams already use. Build and improve RAG systems with a Senior AI Engineer, from chunking and embedding strategies through retrieval quality and evaluation. Manage context windows and model inputs: decide what information a model actually needs, how to structure it, and when a task calls for a different model or a multi-modal approach across text, documents, and images. Help design agentic workflows when the problem calls for it, including multi-step LLM pipelines, tool use, and orchestration. Prompt-engineer and evaluate LLM workflows, treating prompts and retrieval as things you measure rather than guess at. Write clean, testable services and data pipelines that move information reliably between systems. Translate fuzzy workflow needs from investment and operations colleagues into concrete technical problems. Requirements What we're looking for We weigh engineering fundamentals and trajectory far more heavily than years of experience or a specific pedigree. This is a product engineering role: we want a full-stack engineer who is genuinely excited about building with large language models, someone who thinks naturally in terms of retrieval, embeddings, context windows, and tool use, and who wants to ship features real people use day to day. Solid software engineering fundamentals in Node.js and TypeScript: REST APIs, async and promise-based concurrency, a testing framework such as Jest or Vitest, and everyday comfort in git. Our front end runs on AWS Amplify, so familiarity there is a plus. Python is a nice to have, especially for RAG and data-heavy work, but it is not our primary stack. Exposure to cloud infrastructure, ideally AWS, including AWS's AI and LLM services such as Bedrock, Bedrock AgentCore, and Knowledge Bases for Bedrock. Practical experience with LLM APIs (OpenAI, Anthropic, or similar): function calling and tool use, structured outputs, streaming responses, and managing token usage and context window limits. Hands-on experience building something real with LLMs, ideally something you can point to: a project, prototype, or shipped feature you can walk us through, including what broke and how you fixed it. Comfort designing LLM-in-the-loop evaluation and testing pipelines: building golden datasets, using LLM-as-judge scoring, and writing regression tests for prompts and retrieval, with tools such as promptfoo, Braintrust, or Ragas. Comfort with AI native development practices: using tools like Claude Code or Cursor as part of your everyday workflow, not as a novelty. An instinct to measure. When you change a prompt, a retrieval step, or a model, your first question is “how do I know this made it better,” whether the answer is an eval score, a latency number, or a cost-per-call tradeoff. Coachability and initiative: you can take a loosely specified task and make visible progress without needing every detail spelled out. Clear communication. Our users are non-technical investment and operations professionals. Bonus points Working familiarity with RAG mechanics: chunking and embedding strategies, vector stores such as pgvector, Pinecone, or Weaviate, embedding model choice, hybrid search, and reranking. Familiarity with an agent or RAG framework (LangChain.js, LlamaIndex, AWS Strands, or a current agent SDK), though we care more that you can learn one quickly than that you know today's flavor. Experience shipping AI features inside a production full-stack application, not just a standalone prototype. Awareness of data sensitivity: why investor, deal, and LP information cannot flow into arbitrary public models. Exposure to the Model Context Protocol (MCP), whether that's building MCP servers and tools or connecting agents to them. Success metrics By 90 days: you're taking sprint tasks directly, running on a weekly 1:1 cadence with your Senior AI Engineer, and have shipped a medium-significance contribution to an active RAG or agentic workflow project. By one year: you own an increasing share of the day-to-day work on at least one RAG or agentic workflow component, with your Senior AI Engineer retaining design and review oversight, and you're a regular, hands-on contributor to the eval and testing pipeline for the workflows you support. H