01 · Roasts
Serial Sprinter
12 repos, 0 stars, and a graveyard of 2-day wonder-projects: vox-agent died at 9 commits, archLLM-sim at 11, ml-forge at 25. You're great at starting things — finishing them is apparently someone else's problem.
CI? Never Heard of Her
Zero repos with CI out of 12. Not one. You've written LLM orchestrators, Blender agents, and multi-modal OS backends — but apparently a 10-line GitHub Actions YAML is the one engineering challenge that has defeated you.
The License Nihilist
Every single repo: 0 stars, 0 forks, no license. You're building in a legal grey zone for an audience of zero. At least archLLM-sim is consistent — no docs, no tests, no gitignore, maximum ambiguity.
README Roulette
vectra has design.md, ARCHITECTURE.md, STATUS.md, and docs/ — but no README. You wrote four documents to explain a project and skipped the one document GitHub actually shows people. Impressive commitment to being undiscoverable.
Hackathon Hero, Production Zero
echo-OS won a hackathon and has an 80 MB codebase with langgraph loops and multi-modal TTS. It also has 0 stars, 0 forks, no tests, no CI, and no type safety. The judges were impressed; the internet has no idea it exists.
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% weight68C
- Consistency20% weight65C
- Quality20% weight72B
- Depth15% weight65C
- Breadth10% weight55D
- Community10% weight25F
03 · Stats
365-day commit heatmap
85 active days
Language distribution
- Python81%
- PLpgSQL15%
- Cython3%
- C1%
- TypeScript0%
- C++0%
04 · Numbers
Owned repos
non-fork
15
Commits
last 12 months
616
Followers
4
Joined GitHub
Sep 2025
05 · Top repos
v1shay /
substack-network
SQL/Python Substack network crawler with typed language, structured architecture, tests, CI, comprehensive docs (docs/, ARCHITECTURE.md, design.md), and active 3-week development. Complex data pipeline for network analysis with comment enrichment and semantic embeddings.
v1shay /
vectra
Personal experimental project building a Blender AI assistant with agent reasoning, tool registry, and execution engine. Typed Python with comprehensive multi-file architecture, tests, and design docs (docs/ folder, design.md, ARCHITECTURE.md, STATUS.md), but no README, no CI, and no license.
v1shay /
echo-OS
Ambitious voice-first OS for accessibility (hackathon winner); functional agentic backend with langgraph loops, React frontend, and multi-modal STT/TTS integration; lacks tests, CI, and type safety in critical paths despite ~80 MB codebase.
v1shay /
sift
Early-stage TypeScript/Python full-stack app for semantic GitHub exploration via vector DB and LLM-powered search. Typed, documented, multi-layer architecture, but brand new (7 days old) with no users, stars, or external validation.
v1shay /
frontend
Personal portfolio built with Next.js, React, TypeScript, Three.js shaders, and Tailwind CSS. Typed, structured multi-file layout with shader background effects. No README, tests, CI, or license; 22 of 30 recent commits in ~40 days indicates active but recent work. Demonstrates professional frontend craftsmanship but l
v1shay /
neural-lens
Chrome extension + FastAPI backend for real-time in-context text analysis. Typed backend (Pydantic models), documented (docs/ folder, ARCHITECTURE.md, STATUS.md), modular structure, but nascent project with 0 stars and significant placeholder code.
v1shay /
novaSTEM-frontend
TypeScript Next.js 14 nonprofit website frontend with React 18, Tailwind, and Framer Motion animations. Typed, structured multi-file layout (app/, components/, store/), but lacks README, tests, CI, license, and documentation.
v1shay /
ml-forge
Early-stage ML orchestration pipeline with typed Python architecture for data ingestion, profiling, strategy inference, and model training. Recently created with active commits but minimal documentation and no CI/license yet.
v1shay /
upliftArt-frontend
Early-stage TypeScript+React+Vite foundation website with polished 3D animations, shader backgrounds, and responsive design. No tests/CI; thin output but typed and structured, created within weeks.
v1shay /
taquizas-chapala-web
Freelance catering website built in React/TSX with 30 commits over ~16 days. No README, tests, CI, or license; minimal public adoption signals (0 stars/forks). Typed language and structured multi-file layout (6.3 MB codebase) suggest professional work but lack of documentation and test infrastructure limits quality ass
v1shay /
vox-agent
Fresh hardware-integrated speech-to-LLM project (2 days old, 9 commits) with structured Python modules (stream.py, transcribe.py, llm.py, run_live.py), typed code, and clear README. No tests, CI, or license; architectural scope suggests ~500 LOC suggesting one-sprint depth.
v1shay /
archLLM-sim
Bare-bones C++ simulation project with zero ecosystem presence, no documentation, no tests, and minimal file structure. Created 2 days ago with 11 commits—a short sprint with no supporting artifacts for reproducibility or understanding.
06 · Timeline
- Sep 23, 2025Joined GitHub
- Jan 26, 2026Created neural-lens — Real-time in-context data analysis from on-page user selections using a Manifest V3, FastAPI backend, and modular analysis pipeline
- Feb 9, 2026Created taquizas-chapala-web — Freelance development project for local catering service built in TSX, CSS, JavaScript, and React.js
- Feb 10, 2026Created ml-forge — Pipeline + sandbox for data ingestion, AI-powered feature engineering and regression, model development/testing, and graphs/research paper generation
- Feb 21, 2026Created vox-agent — Hardware-integrated Python agent, taking AirPods CoreAudio input and using speech recognition to deliver LLM-powered insight
- Feb 23, 2026Created archLLM-sim — C++ simulation optimizing LLM memory architectures that improves budget adherence by 95% and reduces HBM pressure by 30% via hardware-level token constraints
- Feb 28, 2026Created echo-OS — Voice-first OS for blind users enabling full hands-free computer control via natural language; winner of ElevenLabs-backed hackathon
- Mar 16, 2026Created frontend — Personal website built with TSX, React, Next.js, Three.js, shaders, Tailwind CSS
- Mar 22, 2026Created novaSTEM-frontend — NovaSTEM nonprofit website showcasing programs, impact, and student outreach, built with Next.js 14, React 18, and Tailwind CSS
- Mar 26, 2026Created substack-network — A large-scale Substack network modeling system, developed through research at Chapman, integrating graph construction on a SQL-centric architecture
- Mar 26, 2026Created vectra
- Mar 26, 2026Created upliftArt-frontend — Uplift Art foundation website showcasing mission, impact, and card galleries, built with React 18, GLSL, and Vite
- Apr 18, 2026Created sift — A vectorized representation of GitHub. Query in natural language, let AI sift and surface the perfect repo for you. Ship your next PR in minutes
- Apr 25, 2026Most recent push to sift
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.