Canadian jobs and immigration startup
Job seekers heading to Canada see matched jobs and immigration pathways, scores explained
A Canadian startup wanted international job seekers to see which jobs fit them and which immigration pathways they may qualify for. We built the MVP: AI reads the CV into an editable profile, every match shows its reasons, and pathways come from rules on official data.
B2B SoftwareAI jobs and immigration platformBuilt from zero
The challenge
Its users could not see which Canadian jobs fit them or which immigration pathways they might qualify for, and the requirement documents did not fully agree.
What we did
We wrote one MVP scope from the documents, then built a product that explains itself: an editable AI profile from the CV, scored matches with reasons and rule based pathway indicators.
The outcome
The core job seeker flow was deployed in the client's own Google Cloud account and handed over with its repositories, technical documentation and an ERD.
- Deployed
- In the client's own Google Cloud account
- Match scores
- Each with the top factors behind it
- Pathways
- Rule based indicators with a rationale
- Labour data
- Official open Canadian datasets from day one
The starting point
The client is a Canadian startup building a two-sided platform for international job seekers and Canadian employers. It came with a vision, wireframe concepts and requirement documents, and no existing product.
What had to change
Job seekers struggle to see which jobs fit their skills, which immigration pathways they qualify for and whether there is demand for their profile, and guidance is slow and consultant-led. Employers need candidates who fit both the role and its immigration requirements.
What could not break
- Pathway results stay an eligibility indicator, never legal advice
- Every match score shows the factors behind it
- Accessibility built in from the first design
- Designs frozen before written requirements existed
The calls we made
We narrowed the MVP to what could be explained and trusted, and grounded it in official Canadian data.
- 01
One scope from documents that disagreed
The client's requirement documents did not fully agree, so we compared them, listed what each asked for, and wrote one MVP scope that named what stayed out.
Ruled outBuilding to the widest document and missing the first release.
- 02
Indicators, not verdicts
A rules engine shows a few immigration pathways, each marked eligible, partly eligible or not, with a rationale, so the product never presents itself as legal advice.
Ruled outA full points calculator and full provincial coverage in the MVP.
- 03
Every score shows its reasons
Matching weighs skills, experience, occupation and pathway fit into one score, and the seeker sees the top factors behind it.
- 04
Official data while the API was pending
While access to the live government API waited for approval, we loaded the government's open occupation and labour market datasets, so real data drove the product from day one, with a swap to the live API planned.
Ruled outWaiting for the live API, or building on placeholder data.
- 05
A profile the seeker can correct
AI parses the CV into a profile, and the seeker can edit anything it got wrong before that profile feeds matching and pathways.
How the system fits together
What we built
CV to editable profile
Seekers upload a CV, AI parses it into a profile, and they can edit anything before it feeds matching and pathways.
Explained job matching
Each job match gets a score with the top factors behind it, plus search with filters and job details.
Applications
Seekers apply with a pre-filled form, add a cover letter if they want, and track each application.
Pathways and market view
Pathway indicators with a rationale sit next to salary insights, demand labels from official labour market data and a skills gap view.
How it works
Sign up
The seeker signs up and picks the job seeker role.
Upload a CV
AI parses the CV into a profile the seeker can correct.
See matches
Jobs arrive scored, each with the factors behind its score.
Check pathways
The rules engine shows whether each pathway looks eligible, partly eligible or not, and why.
Apply
The seeker applies with a pre-filled form and tracks the status.
A closer look
What changed
The startup went from a vision and requirement documents that did not fully agree to a deployed MVP with the core job seeker flow: sign up, CV parsing, explained matches, immigration pathways, market insights and applications. It runs in the client's own cloud, handed over with its repositories and documentation.
Under the hood
- Datics role
- Product built by Datics
- Integrations
- A hosted language model, Google and LinkedIn sign in, and a Supabase database for auth and data.
- Data
- Official open occupation and labour market tables feed occupation detection, salary bands and demand labels, cited to the Government of Canada.
- Boundaries
- Pathways are an eligibility indicator with a rationale, not a points calculation or legal advice.
- Deployment
- Review builds on a separate environment during development, then the client's own Google Cloud account, with CI/CD from GitHub.
AI and models
Data and retrieval
Apps and APIs
Also in the stack
How it shipped
Scope
A proposal, then one MVP scope reconciled from the client's requirement documents, with what stayed out listed.
Design
Every screen for job seekers, employers and admins designed first, then turned into a working responsive frontend.
Two build sprints
Sign up and roles first, then CV parsing, matching with reasons, the pathway and market views, search and applications.
Handover
Repositories, technical documentation and a data model diagram, with the MVP deployed into the client's own cloud.
Against the client's backlog
The client's own project manager ran the numbered backlog and sprint reviews, and we delivered against it sprint by sprint.
Scope questions in writing
When a request fell outside the agreed scope, we flagged it in writing before designing it.
Delivered into their cloud
We moved onto the client's own database project during the build and deployed the MVP into its cloud account, after review builds on a separate environment.
What this unlocked
The employer side was next on the client's backlog, with employer screens already built on the frontend.
Related work
Have a product with the same kind of constraints?
Tell us what already exists and what you want it to do. We will tell you what it takes.