01 · Roasts
Hackathon account, not GitHub profile
Of your 8 repos, at least 4 were built in under 3 days each. podagent: 9 commits in 3 days. newteacher: 2 commits in 2 hours. trainmyowngpt: 6 commits in 3 days. You're not building a portfolio — you're leaving hackathon artifacts everywhere.
CI? Never heard of her.
Zero CI pipelines. Zero test suites. Across 8 repos and 9 public projects. Not one. The letters 'CI' don't appear in a single config file. You're shipping blind and calling it bold.
The heatmap has more gaps than a Swiss cheese
Weeks 6 through 16 are completely empty. Weeks 21 through 37 nearly dead. Your 96 yearly commits are scattered across maybe 8 active weeks. That's not building — that's cramming before a deadline.
keeya-py: 6 months, 14 commits
Your longest-running project, keeya-py, has been alive since September 2025. In ~6 months you managed 14 commits, no type hints, no tests, and a README shorter than this roast. v1.0.5 published to PyPI and still no one's using it.
4 followers, 1 following, infinite confidence
You've shipped 8 projects in 13 months, opened 14 PRs, have a live demo at demo.usematcha.dev — and somehow 4 people follow you. One of them might be a bot. You're building in complete silence.
Built using
Zoral
Shadows one worker for a week, then takes over their job with zero extra setup. Behaves exactly like the original.
zoral.ai
02 · Category breakdown
- Impact25% weight62C
- Consistency20% weight55D
- Quality20% weight57D
- Depth15% weight35F
- Breadth10% weight55D
- Community10% weight25F
03 · Stats
365-day commit heatmap
42 active days
Language distribution
- TypeScript51%
- JavaScript32%
- CSS10%
- Python7%
- Shell0%
- Dockerfile0%
04 · Numbers
Owned repos
non-fork
8
Commits
last 12 months
96
Followers
4
Joined GitHub
Feb 2025
05 · Top repos
jayyvk /
trainmyowngpt
Browser-based GPT trainer porting Karpathy's microgpt to JavaScript with Web Workers. Functional interactive tool with dataset presets, real-time loss visualization, and text generation. Early-stage project (3 days old, 6 commits) lacking tests, CI, and license.
jayyvk /
situationtracker
TypeScript Next.js dashboard aggregating geopolitical news from GDELT and RSS feeds with interactive map visualization. Young project (9 days old, 8 commits), typed and structured but lacks tests/CI. Shows solid architectural foundation for real-time situational awareness.
jayyvk /
howaiseesme
Early-stage Next.js project implementing CLIP vision embeddings in browser with real-time webcam visualization. Clean React architecture with Web Worker offloading, typed dependencies, structured layout, and clear README explaining 512-dim embedding visualization and text similarity matching.
jayyvk /
podagent
Fresh Next.js 15 + TypeScript podcast AI agent, built for a hackathon (Firecrawl x ElevenAgents). Integrates ElevenLabs voice agents with live web search via Firecrawl. Well-structured frontend with proper type definitions and API routes, but nascent (9 commits in 3 days), no tests, no CI, and explicit demo-mode gates
jayyvk /
keeya-py
Early-stage AI code-generation library for data analysis using Gemini API. Untyped Python, no tests/CI, but clean prompts, working API integration, and pyproject.toml publishing setup. Personal project with modest scope and ~41KB codebase.
jayyvk /
matcha-playground
Early-stage personal playground for Matcha energy observability product; untyped JavaScript, minimal structure, no tests or CI, but ships a working demo with rich mock data and interactive agent workflow visualization.
jayyvk /
livetax-agent
Specialized Google Gemini Live tax form assistant built for a recent challenge, combining TypeScript frontend (Next.js), Python FastAPI backend, and multimodal AI integration. Functional but experimental, created within days with no ecosystem reach.
jayyvk /
newteacher
Voice-first AI tutoring app integrating ElevenLabs agents, Firecrawl live context, and PDF materials. TypeScript Next.js 15 frontend with reactive React 19. Shipped but minimal adoption (0 stars, 2 commits in ~2 hours).
06 · Timeline
- Feb 20, 2025Joined GitHub
- Sep 21, 2025Created keeya-py — AI powered Python Library
- Feb 26, 2026Created howaiseesme — no matter how advanced AI gets, this is all it sees – numbers.
- Feb 26, 2026Created trainmyowngpt — karpathy's microgpt, ported to JavaScript and running entirely in your browser.
- Mar 1, 2026Created situationtracker — monitoring the situation
- Mar 9, 2026Created matcha-playground — playground to show how energy observability works
- Mar 16, 2026Created livetax-agent — tax assistant using gemini live agent
- Mar 21, 2026Created podagent — ai agent as your podcast guest
- Mar 23, 2026Created newteacher — AI powered Teacher
- Apr 1, 2026Most recent push to matcha-playground
07 · Compare
08 · Rubric
How this score was produced
Overall = Σ (category × weight) + gentle top-end curve
Tier thresholds
▸ How the pipeline works
- 01Scrape.Pull every non-fork repo pushed in the last 90 days, plus your contribution calendar, followers, and language byte counts — straight from GitHub's REST & GraphQL APIs.
- 02Triage.A small model reads every repo's file tree + README and picks the 20 files per repo that actually reveal how you code.
- 03Grade each repo. All repos run in parallel through a fast scoring model that reads the picked files and rates each one independently on Impact, Quality, and Depth — with evidence citations.
- 04Aggregate. A larger reasoning model combines the per-repo scores with server-computed stats (heatmap, commit cadence, language entropy, follower count) to produce the 6-dimension profile score + roasts.
- 05Correct.Deterministic server-side checks enforce anchor-scale floors (e.g. a profile with 2,000+ public commits can't score 30 Consistency) and recompute the final verdict.
~90 seconds per profile, ~$0.25 in compute. Total of ~240 files read across your top-12 repos. One rating per GitHub account per day.
▸ Data sources & caveats
- Heatmap & commit totals: GitHub GraphQL
contributionsCollection— covers the last 365 days, includes private repos when the user has opted in (default). - Language %: byte totals across the top 30 owned non-fork repos.
- Curve: a small upward nudge centered on raw score ≈ 70, capping at 100. Prevents specialists from being unfairly penalised for narrow breadth.
- Anchor corrections: when server-measured signals (e.g. privateWorkLikely, multiRepoVolume, follower count) mandate a minimum category score, the aggregation step enforces it. These are signal-conditional, not identity-based floors.