Dhruv built a thing. Not “plans to build” — built. Shipped, deployed, demoed live in front of a room. This is the distinction that matters before anything else is said about the work itself.
The response has already started arriving. Questions from classmates. People asking to try it. The feedback loop that only opens when something exists to respond to. None of that was possible before DevProof existed. DevProof had to exist first. That is not a small thing. That is, in fact, the whole logic — the only mechanism by which any conversation about this could have occurred.
This is also a first version. It has the properties of a first version: some features more complete than others, some edges still rough, some ideas that will look different in retrospect. That is not a flaw in this product. That is what a first version is. The engineer Dhruv becomes is the one who reads this README in two years and sees exactly what to build next. That engineer does not exist yet. He is being made right now, by having shipped this.
What follows is DevProof — in Dhruv’s own words and his own demo.
DevProof: Prove Your Code. Build Your Credibility.
The problem is real and specific: entry-level developers are competing in a market that has moved past the resume. Companies and startups want to see what you have built. They want commit history. They want merged pull requests. They want evidence of sustained technical judgment — not claims about it.
The credential gap this creates is brutal. Junior engineers have skills but no platform that makes those skills legible to employers quickly. GitHub profiles exist but nobody looks at them. Open source contribution is the most credible path forward, but the issues are buried, the discovery is opaque, and the verification — proof that you actually merged that PR, in that repository, on that date — lives nowhere central.
DevProof is the infrastructure that was missing.
What it does, precisely:
The core is AI-powered semantic search across 10,000+ open source issues. A developer types a natural language query — “beginner Python issues in machine learning projects,” “unassigned TypeScript bugs in popular repos,” “documentation issues I can fix today” — and Gemini AI parses the intent, extracts language, difficulty, labels, and recency, then routes it through Pinecone’s vector database for semantic matching. The ranking weights relevance (40%), freshness (35%), and popularity (25%). The result is not a list of issues. It is the right issue, surfaced in seconds, for the developer who needs a specific entry point.
Once a developer finds and works an issue, DevProof tracks it. When the PR merges, the contribution is cryptographically verified via the GitHub API — authorship confirmed, lines added and removed recorded, merge date logged. This is not a developer claiming they contributed. This is a verified record that they did.
The output is a public portfolio at devproof.io/p/username: verified PR count, lines of code contributed, contribution timeline, shareable to LinkedIn and X in one click. Not claims. Evidence.
The technical architecture:
Next.js 16 frontend on Vercel. FastAPI backend on GCP Cloud Run. Pinecone for vector search. PostgreSQL for user data and contributions. Google Gemini 2.0 for both embedding generation and query parsing. BetterAuth for GitHub OAuth. Issues are ingested on a tiered schedule — every four hours for new issues regardless of repo size, daily deep refreshes for repositories with 100+ stars — so the search index stays current.
The score:
Beyond issue-finding and portfolio building, DevProof evaluates the repositories a developer contributes to across engineering discipline dimensions: cognitive complexity, cyclomatic complexity, commit pattern analysis, engineering standards adherence. The score is not self-reported. It is computed from the commit history. Recommendations accompany the score — specific, actionable guidance for how to push it higher.
Why this matters:
The entry-level developer problem is a discovery problem on both sides. Developers cannot easily find their first meaningful open source contribution. Companies cannot easily find developers whose work is verifiable rather than merely claimed. DevProof addresses both simultaneously — building the portfolio infrastructure that makes junior talent discoverable by the evidence of what they have actually done.
The profile can be made discoverable to companies. DevProof surfaces developers based on what they have worked on. Not keyword-matched resumes. Verified contribution records.
Current status: Live. The demo runs. The search works. The verification pipeline is operational. New features are in development.
Try it: Live Demo · Discord · GitHub — Issue Finder · GitHub — Portfolio & Search · Arkenos World
Built by Dhruv, Northeastern student. Feedback — critical feedback especially — is welcome. Reach out directly.
