Project overview
A more transparent approach to candidate evaluation
Veriloop is a hiring intelligence prototype built during a Tech Thursday hackathon. It explores how AI can help teams evaluate candidates consistently while keeping the evidence and final decision visible to people.
I took the project from concept to working full-stack demo within the hackathon timeframe, then presented the product and its decision-support approach to the judges.
Product approach
AI recommendations grounded in reviewable evidence
- Candidate evaluation against clearly defined job requirements instead of basic keyword matching.
- Requirement-level explanations that help reviewers understand why a candidate received a particular result.
- Structured referral context that adds another source of information without replacing human judgment.
- Constructive candidate feedback that reviewers can edit before making or communicating a decision.
My contribution
Full-stack delivery under hackathon constraints
- Defined the product scope and prioritized the strongest demonstration of the core idea.
- Designed and built the complete user experience and supporting application services.
- Connected document analysis, structured evaluation, and human review into one working flow.
- Prepared the product demonstration and communicated the technical and business value to the judges.
Responsible AI
Designed to support decisions, not make them invisibly
The prototype keeps AI output explainable and reviewable. Recommendations are presented with supporting context, while important hiring decisions and candidate communication remain under human control.
Technical foundation
A modern full-stack TypeScript prototype
Veriloop uses Next.js, React, TypeScript, PostgreSQL, Drizzle, Zod, Tailwind CSS, shadcn/ui, Mastra, and OpenAI tooling. The architecture was deliberately scoped for rapid delivery while keeping product areas modular enough to demonstrate the complete concept.
Result
Second prize at Tech Thursday
Veriloop received second prize at the Tech Thursday hackathon after I presented the working prototype and product rationale.