AI Engineering Intern

About WisBot

WisBot is an AI operator company. We don't sell software — we deploy and run AI agents inside our clients' actual workflows, and we get paid on output: loans screened, deals qualified, briefs produced. Our clients are in private markets and financial services, including a commercial real estate lending platform where our agents run live origination workflows every day.

Behind those deployments is the WisBot platform — the internal system that lets us stand up, run, and monitor agent operations for each client. It's the engine of the company, and it's actively being built.

We're a small team. Interns here ship to production, not to a sandbox.

What you'll do

You'll work across both sides of the company:

Ship to production

• Build and iterate on LLM agents and multi-step workflows that run live in client environments, processing real deals and real documents
• Own a feature end to end — scoped with the founder, reviewed by founding engineers, deployed, then monitored while real users depend on it
• Handle the unglamorous 80%: document parsing, data extraction, schema design, prompt iteration, error handling on messy real-world inputs
• Respond to what production actually surfaces — an agent misreads a rent roll, a job hangs, a client flags an output — and fix it

Build the WisBot platform

• Contribute directly to the platform that powers every deployment: the job execution layer, agent orchestration, the developer portal, and client-facing screens
• Build the durable job queue, retry logic, and observability that keep agent operations running reliably at scale
• Write evaluation harnesses and eval tooling into the platform so agent quality is measurable across every client, not just tested by hand
• Turn one-off deployment work into reusable platform capability — this is the core engineering discipline here, and it's how the company gets leverage

What you'll learn

Most AI internships end with a demo nobody uses. This one is built so you leave with things you can point to:

• What it means to ship AI to production. Code you write will run against real client work with real money attached. You'll learn deployment, monitoring, incident response, and the discipline of writing systems that fail safely — not just code that works on your machine.
• How to build platform, not one-offs. The hardest problem in applied AI services is turning bespoke work into reusable infrastructure. You'll be doing exactly that, and you'll learn to recognize the difference in your own designs.
• How agents actually behave in production. Not benchmark performance — real failure modes on real documents, and the engineering patterns that contain them: retries, fallbacks, human-in-the-loop checkpoints, graceful degradation.
• Evaluation as a discipline. You'll build eval sets from real client data and learn to measure agent quality when there's no clean ground truth. The single most transferable skill in applied AI right now, and almost nobody teaches it formally.
• The domain underneath the model. How credit underwriting, loan origination, and investment diligence actually work — the workflows, the documents, the decisions. Financial services domain knowledge plus AI engineering is a rare and durable combination.
• How AI work gets priced and sold. You'll see how we scope a deployment, define the output we're accountable for, and price it. Rare exposure for an engineering intern, and useful whether you go into industry or start something yourself.
• A portfolio you can actually talk about. You'll leave owning a component of a live client deployment and a piece of the WisBot platform — concrete work with concrete results, not a slide. Strong interns are considered for a return offer or full-time role.

Minimum requirements

• Currently pursuing a BS/MS in Computer Science, Engineering, Data Science, or a related field
• Strong Python; comfortable reading and extending someone else's codebase
• Hands-on experience building something with an LLM API (course project, side project, hackathon, or prior internship all count)
• Familiarity with Git and basic backend concepts (APIs, async, databases)
• Comfortable with the pace and ambiguity of a small team shipping to production
• Able to communicate clearly in writing — you'll be documenting decisions, not just writing code

Nice to have

• Experience with agent frameworks, RAG pipelines, or structured output/tool use
• Backend infrastructure experience: queues, workers, Docker, async job processing
• Front-end skills (React/TypeScript) for platform and client-facing screens
• Exposure to financial services, lending, or investment workflows
• Any experience with evals, observability, or prompt versioning