01 · Roasts
The 5-Minute Repo
dl_fit was created and last pushed within 5 minutes on 2026-02-12. That's not a project, that's a README with ambition and commitment issues.
Contest Gamer, Not Open Source Contributor
Your most-starred repo (cmimc26, 2 stars) is a 3-day contest sprint with a README that just says 'Team name: copypaste'. The open-source community has 4 followers' worth of opinions about this.
Architecture Docs > Actual Architecture
openrayv2 has ARCHITECTURE.md, design.md, AND STATUS.md at 3 days old. Meanwhile your portfolio site — your actual public face — has no README at all. Priorities are… interesting.
84% Solo, 4 Followers, 0 PRs Merged
With 84% solo commits, 4 total PRs this year, and 4 followers, your GitHub is technically a distributed system where you are every node. Very decentralized. Very lonely.
Sprint Merchant
glyde (5 days), openrayv2 (3 days), interactive-transformer (1 day), dl_fit (5 minutes) — you ship fast but none of these projects have been touched since their birth week. LuxenAI's graveyard awaits.
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% weight56D
- Consistency20% weight65C
- Quality20% weight57D
- Depth15% weight58D
- Breadth10% weight65C
- Community10% weight25F
03 · Stats
365-day commit heatmap
122 active days
Language distribution
- Python48%
- TypeScript32%
- JavaScript8%
- Swift4%
- HTML3%
- Cython1%
- Other4%
04 · Numbers
Owned repos
non-fork
55
Commits
last 12 months
371
Followers
4
Joined GitHub
Jun 2021
05 · Top repos
g4nesh /
luxenllc
Marketing website for Luxen LLC's AgentTree AI orchestration platform, built with Next.js and TypeScript. Typed, documented with README, structured multi-file layout. Exports as static site with interactive canvas animations. No tests or CI, but clear business product. Active but still early-stage startup presence.
g4nesh /
parser
TypeScript DOM-based MCTS agent framework for computer-use tasks, demonstrating focused specialization in planning with clean architecture and type discipline despite minimal adoption signals.
g4nesh /
openrayv2
Early-stage TypeScript MCTS web agent project (1 star, 3 days old) with comprehensive docs, tests, and typed code. Novel browser automation approach but pre-production maturity.
g4nesh /
interactive-transformer
Browser-based interactive transformer visualization using Hugging Face's DistilGPT2 with Three.js 3D rendering. Novel educational tool showing token-by-token generation and layer activity, but nascent (13 commits in 1 day, no tests/CI).
g4nesh /
cmimc26
Competitive programming contest submission (CMIMC 2026) with multi-domain solver implementations for maze navigation, image recovery, and computational puzzles. Sparse documentation, no tests/CI.
g4nesh /
glyde
Early-stage glycan ML research project with typed Python encoder for domain-aware embeddings. Has structured code (encoder.py, main.py) and basic documentation, but minimal adoption (1 star, 0 forks), no tests/CI, and limited git history (5 days old, 11 of last 30 commits).
g4nesh /
portfolio
Personal portfolio website built in HTML with 14.8 MB codebase, 30 commits over 7+ months, but lacks README, documentation, tests, CI, license, and gitignore—minimal project structure and no evidence of architectural depth.
g4nesh /
dl_fit
Standalone bone segmentation tool with README describing 2D/2.5D/3D training and inference, but extremely early-stage with 1 star, 2 recent commits in 5 minutes, no tests, CI, license, or visible source code.
g4nesh /
g4nesh
Nearly-empty repository with only a minimal README redirecting to author's portfolio site. No source files, tests, CI, or documentation beyond a single redirect link.
06 · Timeline
- Jun 30, 2021Joined GitHub
- Sep 16, 2025Created portfolio — https://ganeshtalluri.com
- Sep 16, 2025Created luxenllc — luxen llc website
- Nov 9, 2025Created g4nesh — Github README
- Feb 8, 2026Created glyde — A domain-aware, topology-biased glycan language model for viral receptor binding
- Feb 10, 2026Created parser — Monte Carlo Tree Search (MCTS) adjacent DOM for low-latency tool-assisted computer-use agents
- Feb 12, 2026Created dl_fit — 2D Software for Bone Segmentation
- Feb 18, 2026Created openrayv2 — OpenClaw on steroids. Boost web browser inference and wall clock time with the OpenRay extension.
- Mar 12, 2026Created interactive-transformer — Interactive transformer model. Inspired by Karpathy's ConvNetJS series (https://cs.stanford.edu/people/karpathy/convnetjs/)
- Apr 17, 2026Created cmimc26
- Apr 27, 2026Most recent push to portfolio
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.