AI Engineering Intern
Build features on top of large language models: retrieval, tool use, evaluation, and the unglamorous plumbing that makes a demo into something a client can rely on.
Apply for this role- Location
- Remote, with Regina, SK as home base
- Commitment
- Part-time during term, full-time over the summer
- Intake
- Summer 2027 intake, rolling review
- Team
- AI & Applied Machine Learning
- Level
- Student or recent graduate
About the role
Most of the AI work we ship is not model training. It is the layer around a model: getting the right context in front of it, giving it tools it can actually call, measuring whether the answer was any good, and failing safely when it was not.
You will work on that layer. Expect to spend as much time on evaluation harnesses and data plumbing as on prompts, because that is the part that decides whether a feature survives contact with real users.
You do not need prior machine learning experience. You do need to be comfortable in Python and genuinely curious about why a system produced the output it produced.
What you will actually do
- Build and iterate on retrieval pipelines: chunking, embedding, indexing, and the ranking step everyone forgets about.
- Wire models to real tools and APIs, and handle the cases where the model calls them wrong.
- Write evaluation sets for features you build, so 'it got better' is a number rather than a feeling.
- Trace and reduce latency and token cost on live features.
- Write up what you found, in language a client's operations lead can follow.
What you need on day one
- Comfortable writing Python: functions, classes, virtual environments, reading a stack trace without panic.
- You have called an HTTP API from code and handled its errors.
- You can read documentation and try things before asking, and you ask clearly when you do.
- Enough written English to explain a technical decision to someone who was not there.
Nice to have, genuinely optional
None of these are requirements. They are listed so you know what would be a head start, not so you can talk yourself out of applying. Apply if the list above fits.
- Anything you have built with an LLM API, however small or scrappy.
- Familiarity with vector databases, or with the maths behind embeddings.
- Coursework in statistics, linear algebra, or information retrieval.
- Git beyond `commit` and `push`: branches, rebases, reviewing someone else's diff.
What you should expect to learn
- How production AI features are actually structured, past the tutorial layer.
- How to measure a non-deterministic system, which is a skill most people never get taught.
- How to scope AI work for a client who has a budget and a deadline.