The company makes an AI video product that helps people turn longer footage into short clips for social media. It launched 18 months ago and more than 20 million creators and businesses have used it.
The company has raised $50 million, is headquartered in Mountain View, and has a team of 100.
This co-op sits on the Customer Experience team and turns real support tickets into automated end-to-end tests of the user interface. You will be the first person on what the team calls customer-originated requirements. Each month's thousands of support conversations show how the product is really used and where it breaks, but nobody has yet turned that into a repeatable set of tests.
The team moves people from CX into data, engineering and customer-facing technical roles, and this internship is one route into that path.
Responsibilities
Build a UI end-to-end test battery from real customer reports. Review support chats and bug tickets for the tasks customers perform and the points where they run into trouble, reproduce those failures, and write tests that catch them. Link every test to its source ticket so coverage follows actual user problems. Results are judged by how many customer-reported regressions your tests would have caught and how quickly CX can confirm a reported issue is genuine.
Break user flows into testable steps. For a flow such as uploading a video, getting clips and exporting to TikTok, identify each step, each user choice and each failure point, then rank them by how many customers they affect. Do this with the CX team, who hold the ticket data, and document it so Product and Engineering can reuse it as an ongoing map of product use.
Work AI-native. Use Claude Code or Codex daily: define the task, supply context, split the work into steps, check the output and catch wrong answers. Help CX adopt the same habits by sharing patterns and small tools that let non-engineers verify issues and query data themselves.
Connect CX and Engineering. Your work lives in Engineering's repositories and CI pipeline. Open pull requests, respond to code review, follow branch and merge rules, and report flaky or failing tests with clear reproduction steps. When a test reveals a product gap, write up the customer context, business impact and suggested priority so the owners can act without follow-up questions.
Requirements
Must-have: You have built and shipped something real with AI help, such as a side project, internal tool or script, and can describe what you requested, what the model got wrong and how you corrected it.
Must-have: You can follow a short piece of Python or JavaScript/TypeScript and notice a clear error. Writing either language from scratch is not required.
Must-have: Given a feature, you list the normal path, edge cases and failure modes, then rank them by user impact instead of by how easy they are to test.
Must-have: You want to understand what the user was attempting, not just what broke, and you see the person behind a ticket.
Must-have: You write short, specific reports that someone who was not present can follow, and you request and accept feedback smoothly.
Must-have: You can weigh return on effort and opportunity cost, such as why a bug affecting 2% of paid exports matters more than one in a rarely used free feature, and what testing the wrong thing costs.
Must-have: You can split a flow like uploading a video, getting clips and posting to TikTok into each step, each user decision and each point where it can fail.
Nice-to-have: You have made, edited or published video, or follow creators closely enough to judge a good clip and why people pay for one.
Nice-to-have: You are skilled with Claude Code or Codex, supplying good context, breaking tasks into steps and refining output rather than accepting the first answer.
Nice-to-have: You are comfortable with browser DevTools (Network and Console tabs, element selectors) and with Git branches, commits and pull requests, including responding to review.
Conditions
Four-month co-op based in Vancouver, starting January 7, 2027.
An extension of 8 months is possible, decided jointly rather than assumed. The first 4 months cover a defined backlog that is worthwhile on its own.
Benefits
You own a real test battery that protects paying customers, and your name is on every pull request.
You use AI coding tools every day as part of the job, shipping automation into a live product and building a record of tests tied to customer evidence.
You get direct access to the CX, Product and Engineering teams, and your manager's stated goal is to move strong CX people into technical roles across the company.