Choosing a vehicle
Drivewise is an MVP for comparing vehicles against a buyer’s budget, intended use, and priorities. It brings a vehicle catalog, listings, and supporting documents into a web application backed by a Python API.
Recommendations and evidence
The recommendation flow uses deterministic scoring over reviewed catalog offers. Metric-level provenance records where the underlying information comes from. Model analysis exposes price, maintenance, ownership-cost, and red-flag fields through a defined API contract.
Filters live in page URLs, so searches and filtered lists can be shared and restored through browser navigation. Tests cover API behavior, catalog handling, scoring, and frontend routes.
Current scope
The public repository documents a working MVP foundation. Live crawling and production vector search are deferred; the documented vector mode uses deterministic fake embeddings for development and testing.
My role
I have worked on Drivewise since 2026 as a full-stack and AI developer, focusing on the backend while my co-founder develops the frontend. I define requirements and design the data model, with code reviews shared across the team.