Engineering · Full-time · Toronto, in person five days a week · CAD $68,000 base + equity

15Rock builds research software for advisory, investment, corporate strategy, and climate teams. Our products turn filings, transcripts, PDFs, private data, and client methodologies into source-linked shortlists, Value Scans, benchmarks, and client-ready reports.

We think most of what passes for financial research is unverifiable, and that this is a solvable engineering problem. Every claim our systems produce traces back to a source. That standard is the company.

About the role

Most of the code in our products is written by AI. We say that plainly because it changes what the job is, and you should know before you apply.

Nobody here rations tokens or waits on a seat licence. The typing is fast. The hard part moved somewhere else: deciding what should be built, specifying it precisely enough that a model cannot misread it, reading code you did not write and knowing what is wrong with it, and owning the output once a partner is presenting it to an investment committee.

Very few companies are hiring juniors into that job. Most AI-native startups skipped the junior cohort and hired seniors instead. We think that is a mistake. We would rather find someone early and train them into it.

The problems you'll work on

- When a pipeline claims a company's margins are compressing, how do you verify — automatically, at scale — that the filing it cites actually says that?
- How do you build an evaluation that catches an analysis being subtly wrong before a partner does?
- A rigorous company scan runs long. A board reads three pages. How do you compress the report without breaking the chain of evidence?




- How do you benchmark a company against its peers product by product, when the evidence is scattered across hundreds of filings, transcripts, and datasets?

None of these are solved. They are the job.

What you'll do

- Take a scoped problem from a vague request to something running in production, and stay with it after it ships
- Review generated code and catch what is wrong: logic that is subtly off, failures that stay quiet, output that looks right and isn't
- Build and extend the tests and evaluations that let us trust the pipeline instead of hoping
- Work on citation quality, source traceability, and validation, the machinery that proves a claim in a report is real
- Build the interfaces people use to research, review, and deliver
- Investigate production issues from logs and turn them into reproducible test cases
- One year of professional software experience, or a completed co-op or internship where you shipped something real
- Working fluency in either TypeScript and React, or Python
- A link to something you built that we can look at

That's the whole list. No degree requirement.

You may be a fit if

- You can read code and tell whether it is correct, not just whether it runs
- You can say how you would know something works. We weight this most heavily.
- You write clearly. A vague ticket produces vague code here, so writing is part of the engineering.




- You ask a question before you have been stuck for a day
- You have used AI coding tools seriously enough to know where they fail
- Useful but not expected: retrieval and search systems, evaluations, cloud deployment, data pipelines, B2B analytics and research products

Two things to know before you apply

This is an in-person job, five days a week in our Toronto office. The reason is simple: this work is learned by sitting close to people who are good at it, and that does not travel over video. If you want remote or hybrid, this is not the role.

The work is scrutinised. A partner or an investment committee reads what you build, and when something is wrong they say so, by name, in a meeting. That is uncomfortable. It is also the shortest feedback loop you will find this early in a career.

Compensation and advantages

- Equity under the company's equity incentive plan. We will walk you through exactly what your grant means before you sign anything.
- Health and dental coverage
- RRSP matching
- AI tooling without limits. Frontier models and agentic coding tools, no seat caps, no token budget to manage.
- Small enough that you are in the room for architecture and product decisions, and close enough to delivery that you hear what a client said about the thing you built

Our stack

TypeScript, React, Next.js · Python, FastAPI · PostgreSQL · GCP and Cloud Run · search and language-model APIs · pytest, Vitest, Playwright · GitHub Actions

15Rock welcomes applicants from different backgrounds and learning paths. If you need an accommodation during the hiring process, please tell us what would help.

#J-18808-Ljbffr

📌 Junior Full-Stack Software Developer (AI Products) (Toronto)
🏢 15Rock
📍 Toronto