AI recommendations at production scale · 2025-2026
Govgrant.ca
A Canadian grant-discovery platform that matches businesses with relevant funding programs using RAG-based recommendations.
2,000+ users and 300+ grants refreshed daily
The problem
Canadian businesses had to search fragmented government sources, interpret eligibility criteria, and repeatedly check whether programs were still open. The product needed to make discovery faster without presenting stale or irrelevant opportunities.
My role and scope
I led engineering across the recommendation experience and SaaS platform, turning the product requirements into a reliable matching workflow with authentication, subscriptions, admin controls, and an automated grant pipeline.
Key product decisions
- Combined structured grant data with retrieval-based recommendations so matching could remain explainable and grounded in current program information.
- Automated grant ingestion and refreshes instead of relying on a manually maintained catalogue.
- Treated freshness, eligibility context, and operational admin tools as product requirements, not background infrastructure.
- Built the matching experience inside a complete subscription product rather than shipping a standalone AI demo.
What shipped
- RAG-based matching and recommendation workflows
- Automated ingestion processing more than 300 grants per day
- Authentication, billing, subscriptions, and administration
- Production infrastructure supporting more than 2,000 users
Outcomes
- Served more than 2,000 users across Canada
- Kept the grant catalogue current through a daily automated pipeline
- Connected AI recommendations to a usable, monetizable SaaS workflow
What this demonstrates
- AI product judgment
- Platform thinking
- Reliability and data freshness
- Product delivery at scale