About the role#
We are seeking interns to join a small pod building an enterprise AI capability that helps leadership move from information to insight. This system retrieves from document collections, queries structured data, and calls live data sources. It must be accurate, traceable, and controlled. You will work on the components that ensure trust, including agent design, retrieval, grounding, citation, evaluation, and guardrails.
What you'll need#
- Currently enrolled in a Bachelor's, Master's, or PhD program in Computer Science, AI/Machine Learning, Data Science, Software Engineering, or a related field.
- Strong Python skills and hands-on experience with LLM APIs or frameworks such as Azure OpenAI, Semantic Kernel, LangChain/LangGraph, or LlamaIndex.
- Practical understanding of RAG, embeddings, vector search, and LLM evaluation methods.
- Awareness of LLM reliability and security risks including hallucination, prompt injection, and data leakage.
- Familiarity with Git, testing, APIs, and basic CI/CD.
- Experience with multi-agent or tool-calling architectures, or Azure AI Search, Microsoft Foundry, or Copilot Studio agents.
- Coursework or interest in AI governance, Responsible AI, model risk, regulation, or technology law.
- Experience with financial, market, or news data, or complex legal or financial documents is preferred.
- Ability to bridge business and technology teams, explain technical work in business terms, and turn business needs into technical requirements.
- Experience presenting to senior leaders or executive audiences.
What you'll do#
- Document business processes, decision rules, and handoffs with stakeholders, then translate them into agent steps, tools, and approval points within an orchestrated multi-agent workflow.
- Design and prototype agent orchestration, tool use, retrieval, and prompts on approved enterprise AI platforms.
- Experiment with chunking, embeddings, hybrid search, re-ranking, and metadata filtering to ensure every answer cites its source and shows its currency.
- Build question sets from real executive inquiries with known-good answers to measure accuracy, groundedness, citation fidelity, and hallucination rates.
- Track answer quality, latency, and consumption over time to recommend tuning.
- Prepare inputs for architecture review, Responsible AI and model risk validation, records retention, e-discovery, operational resilience, and audit readiness.
- Provide regular progress updates to senior leadership, adjusting the level of detail for executive and technical audiences.
- Turn governance and risk requirements into design patterns, such as human approval points and access boundaries.
- Write technical documentation and decision records while partnering with data and cybersecurity interns on secure connectors and test plans.
Location & details#
- Term: Winter 2027 (January to June).
- Availability: At least 30 hours per week.
- Modality: Hybrid, with a requirement to work onsite 4 days per week.
- Locations: Burlington, Massachusetts or Quincy, Massachusetts.


