AI Engineer Intern — AI Foundry & Digital Growth

MAIA is Prophet's next leap in AI — an always-available assistant that amplifies human insight. By integrating directly into our workflows, it turns AI into true Augmented Intelligence, extending our creativity, judgment, and expertise. Built on a multi-agent framework, MAIA connects specialized tools, data sources, and reasoning systems through one interface, so our people can do more, think deeper, and deliver sharper outcomes for clients.

As an AI Engineer Intern on the AI Foundry team, you'll sit inside the group building and evolving MAIA, Prophet's multi-agent AI platform, and write code that ships. This is a hands-on technical role, not a rotational program: you'll build and test agents, run evaluation protocols, contribute to client-facing applications on the platform, and take scoped problems from spec to working solution with guidance and code review from Foundry engineers.

This role sits at the intersection of technical and non-technical worlds. Most of your day-to-day colleagues at Prophet come from consulting, strategy, and design backgrounds and have learned AI on the job; the deeper technical counterparts — engineers and platform developers — sit with our external technology partners. You'll get real exposure to both, and you'll often be one of the more technical people in the room.

• Agent Development & Testing — Build and configure agents against defined specifications: system prompts, tool definitions, and data flows. Run iterative prompt engineering and testing cycles to improve accuracy, reliability, and user experience.
• Experimentation & Evaluation — Support experiments that validate new agent capabilities. Build test sets, run scaled testing protocols, and measure accuracy, cost, and latency so results turn into evidence rather than anecdotes.
• Platform & Application Development — Contribute to MAIA and to client-facing applications built on it. Write and debug production-quality code, submit it for review, and act on feedback from testing cycles.
• Technical Research & Benchmarking — Investigate specific technical questions the team is weighing — retrieval strategies, API integration patterns, knowledge graph structures — and document what you find, including tradeoffs across cost, speed, accuracy, and maintainability.
• Instrumentation & Reporting — Help track product and system metrics (adoption, usage, latency, error rates) and pull them into executive summaries.
• Tooling & Enablement — Improve the prompts, templates, evals, and internal workflows that make Foundry tools more useful to the wider firm, and document what you build so the next person can pick it up.
• AI Landscape Monitoring — Stay current on LLMs, agent frameworks, and AI tooling, and surface releases or techniques that could improve MAIA or how the team works.

• Recently completed a Bachelor's degree in Computer Science, Machine Learning, Data Science, Engineering, or an equivalent technical background.
• Demonstrated building experience through coursework, personal or open-source projects, research, or a prior internship — we care more about what you've shipped than how long you've been at it.
• Working Python and SQL, plus Git-based workflows (branches, pull requests, code review) and comfort in an editor like VS Code.
• Hands-on experience building at least one LLM application — RAG (chunking, embeddings, retrieval) or an agent framework such as LangChain, LangGraph, or LlamaIndex — with real prompt engineering behind it.
• Enough full-stack ability to make something usable end to end (e.g., FastAPI or Flask with a React front end, deployed on Vercel or Supabase) and to design and consume REST APIs.
• Practical data or ML exposure — Pandas/NumPy, plus PyTorch or scikit-learn — is a strong plus, as is any exposure to containerization and cloud (Docker, AWS/GCP/Azure).
• The instinct to check your own work — measuring whether output is accurate, fast, and affordable, not just whether it runs.
• Structured analytical thinking and the ability to break a scoped, ambiguous problem into steps