{"id":46981450,"title":"Show HN: DevProof – Verified developer portfolios using Code Complexity and AI","points":2,"user":"dhruv0206","time":1770845892,"time_ago":"9 hours ago","type":"link","content":"

Hi HN,

I'm Dhruv. I built DevProof because I was tired of guessing which "Senior Engineer" actually understood distributed (or complex) systems versus who just pasted code from tutorials. GitHub activity charts (green squares) measure effort, not engineering depth.

What it is: DevProof is a portfolio generator that audits your actual code. It follows a "Proof of Work" model: it pulls your source code, runs it through a 2-step verification pipeline (AI + Static Analysis), and builds a verified profile.

How it works (The Ranking Algo):

Ingestion: We pull your top repos (filtering out forks/docs).\nReasoning (Gemini 3.0 Pro): An AI agent classifies your code into 3 Tiers (Tier 1: UI, Tier 2: Logic, Tier 3: Complex Algos).\nVerification (Static Analysis): We run McCabe Cyclomatic Complexity (using lizard) on the flagged files. If the AI claims a file is "Tier 3" (Complex) but the complexity is < 15, we automatically downgrade it. (This prevents the "hallucination" problem where AI thinks a long file is complex when it's just verbose).\nThe Hiring Engine (Coming Soon): Portfolios are step one. We are building a hiring engine where you search for experience, not keywords.

Example Query: "Find a dev who has extensive experience building Voice AI pipelines."

Instead of sending you to GitHub to dig through repos, we surface the evidence directly:

The Tier: (e.g., Tier 3 - Real-time Audio Processing)\nThe Project: (e.g., VoxArena)\nThe Evidence: Direct links to the specific lines of code and complexity analytics that prove the skill.\nFuture Roadmap: We are also building an AI Detection System to verify human authorship and an Intent Analysis Engine to understand why a developer made specific architectural choices, moving beyond just "what" they wrote.

The Ask: I’d love for you to try it on your own public repos.

Does the "Complexity Score" feel fair?\nDid it correctly identify your "hardest" code vs. your "boilerplate"?\nI am tuning the thresholds for the Tier downgrades right now, so any examples of false positives/negatives would be super helpful.

Dhruv","url":"https://dev-proof-portfolio.vercel.app","domain":"dev-proof-portfolio.vercel.app","comments":[],"comments_count":0}