About Us

DaoAI is a Vancouver-based AI company building applied AI products across two lines of business:

- DaoAI Robotics — AI-powered optical inspection systems for electronics manufacturing
- Wemio — our AI-native video creation platform, which turns scripts and ideas into finished short-form video using a multi-agent production pipeline

This role sits on the Wemio team. You may see this posting listed under DaoAI on your school's co-op portal — same company, and Wemio is what you'll be working on day to day. No manufacturing or robotics background needed. About This Role The role has two halves, and they use the same skill pointed in different directions.

Half one: make what we have better.

Our platform generates video. Sometimes the output is genuinely good. Sometimes something is subtly off — a mouth that doesn't quite match the audio, a blink at the wrong moment, a cut that kills the pacing, lighting that doesn't sit right on a product.

Right now when that happens, the feedback we get is "this one feels wrong." That isn't something engineering can act on. You'll close that gap: run our pipelines against a consistent test set, identify what specifically breaks, isolate which variable caused it, and hand engineering a reproducible case they can fix.

Half two: find out what we could build next.

With generative video, whether a feature is even possible is an empirical question. We can't spec it and hand it to engineering — someone has to try it forty times first and find out where the ceiling is.

That's you. We'll give you a capability we're considering — a new kind of shot, a camera behaviour, a style we want to hold consistently across a whole sequence — and you'll go find out whether it's achievable, what it takes, and where it falls apart. If you get it working by hand, engineering knows what to automate. If it can't be done yet, we've saved a quarter of build time finding out.

Both halves run on the same method: change one thing, compare versions, write down what happened. It's closer to VFX dailies than to content production — a lot of versions, side by side, working out what changed and why.

What You'll Do

Build the test set (your first month)

- Create a standard set of test cases across our three output types — spokesperson video,



product/ad content, and narrative video
- Define what "acceptable" looks like for each, with reference examples
- Establish a repeatable process for running the full set against any pipeline change

Run and compare (ongoing)

- Generate output against the test set every time a model, prompt, or pipeline component changes
- Compare versions systematically and document what improved, what regressed, and what stayed broken
- Maintain version history so we can trace when a problem appeared

Diagnose

- When output is wrong, isolate the cause — change one variable at a time and find which one moves the result
- Write it up so an engineer can reproduce it: what you did, what you expected, what happened, what you already ruled out
- Track recurring failure patterns across many generations, not just one-offs

Explore

- Take a capability we're considering building and find out empirically whether it's achievable — try it, many times, with diverse approaches
- Push past the obvious attempts. Most of what's interesting here sits behind twenty failed tries.
- When something works, document the recipe precisely enough that engineering can automate it
- When something doesn't, document why, what you ruled out, and what would need to change — a well-documented dead end is a real result here, not a failure
- Keep an eye on new models, tools, and techniques as they appear, and tell us which ones are actually worth our time

Report

- Weekly summary of current quality state per output type
- Flag regressions immediately
- Present side-by-side comparisons to the product and engineering team
- Short write-up per exploration: what you were testing, what you tried, what happened, what you'd try next

What We're Looking For Required

- A demo reel or portfolio showing video, animation, VFX, or motion work




- A trained eye — you can look at a shot and say specifically why it isn't working, not just that it isn't
- Patience for methodical, repetitive comparison work. This role involves watching a lot of near-identical clips and spotting small differences. If that sounds tedious rather than satisfying, it isn't the right fit.

- Tolerance for dead ends. On the exploration side, most attempts won't work. You need to be the kind of person who finds the twenty-first attempt interesting rather than demoralizing.
- Self-direction.

We'll tell you what we're trying to find out, not how to find it out.

- Ability to write clearly and precisely — most of your output is written documentation
- Comfort with AI generation tools, or genuine curiosity about them

- Enrolled in a co-op or internship program (Animation, VFX, Film, Motion Picture Arts, Digital Media, or related), and eligible to work in Canada

Nice to Have

- Bilingual English / Mandarin — a real plus. Some of our narrative content and reference material is in Chinese, and it opens up more of the role. Day-to-day work is in English, and English-only candidates are welcome.

- Compositing experience (Nuke, After Effects) — the variable-isolation habit transfers directly
- Character animation background — especially useful for spokesperson output, where lip sync, blink timing, and micro-expression are where things fall apart
- Colour grading or lighting experience

- Any scripting (Python, JavaScript) for batching repetitive generation runs
- Familiarity with prompt-based generation tools and how phrasing changes output

Not required

- A CS degree. We have engineers. What we need is someone who can tell them precisely what's wrong.

What You'll Get
- Hands-on time with the current generation of AI video tooling, well ahead of where most studios are

- Real influence on what we build — the exploration side of this role directly shapes our roadmap
- A rare vantage point: you'll spend months finding out what generative video can and can't actually do, in far more depth than anyone gets from casual use
- Close work with both the creative and engineering sides

- A body of documented work you can point to

Pay: $3,500.00 per month Work Location: Hybrid remote in Vancouver, BC

📌 AI Video R&D Co-op — Output Quality & New Capability Exploration (Intern) (Vancouver)
🏢 DaoAI Robotics
📍 Vancouver