01 · Roasts
Hackathon Sprinter, Not a Finisher
HerdSignal: 3-day sprint. ChhayaAI: 12 days. mhosigiri portfolio: 1 day. Google_buildWithAI: 18 days. You're great at starting things fast — you just never come back to add tests or CI after the adrenaline fades.
CI/CD? Never Heard of Her
Out of 7 analyzed repos, exactly 1 has CI — and that's a fork where upstream wrote it for you. Zero tests, zero pipelines, zero automation on anything you actually built. Your README game is strong; your test suite is imaginary.
156 Commits, 41% Abandoned
You pushed 156 commits this year across burst windows, but 41% of your repos haven't been touched in 2+ years. The heatmap looks like a meteor shower — brief intense impact, then weeks of silence.
8 Stars, 8 Followers, 8 Following
Suspiciously symmetric social metrics. Your most-starred repo has 5 stars (hackathon), your flagship AI workshop has 0. You're building Google Cloud–scale infrastructure and telling roughly nobody about it.
TODOs in Production (Almost)
ChhayaAI's map_agent.py and alert_agent.py have _execute_query and _save_alert stubbed as TODOs. You built a multi-agent emergency response system and left the 'save the alert' part as homework.
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% weight67C
- Depth15% weight65C
- Breadth10% weight65C
- Community10% weight25F
03 · Stats
365-day commit heatmap
46 active days
Language distribution
- TypeScript54%
- C++29%
- C6%
- Vue5%
- Python2%
- C#1%
- Other3%
04 · Numbers
Owned repos
non-fork
37
Commits
last 12 months
156
Followers
8
Joined GitHub
Aug 2021
05 · Top repos
mhosigiri /
Google_buildWithAI
Workshop platform teaching Google Cloud AI via narrative game. Multi-level agent architecture (ADK, MCP, FastMCP) with Next.js 3D frontend, FastAPI backends, cloud infrastructure. 116MB codebase with structured levels but limited adoption and no tests.
mhosigiri /
HerdSignal
HackSMU hackathon full-stack elephant conservation app combining Python NMF audio separation, Next.js frontend, and PostGIS geospatial heatmaps. Typed, documented, multi-file architecture with proven 0.72 match score on 212 field recordings. Recently created (3 days old), not yet adopted beyond the team.
mhosigiri /
gemini-alert-app
Crisis de-escalation app combining Vue 3 frontend with Python/Flask backend, integrating Groq AI for guidance and Firebase for real-time location/alerts. Personal project with typed Python backend and structured architecture but minimal adoption and no tests.
mhosigiri /
ChhayaAI
Ambitious emergency-response iOS app with multi-agent FastAPI backend and Firestore integration. Typed Swift + Python, structured codebase, but nascent adoption (2 stars, 12 days old), minimal tests, lacks CI/CD, and production integrations are stubbed.
mhosigiri /
n8n_local_cluster
Fork of n8n workflow automation platform for extracting viral content from videos. TypeScript codebase (196MB) with README, tests, CI, and architecture docs, but 0 stars/forks and explicitly described as a personal fork limits impact assessment.
mhosigiri /
mhosigiri
Personal portfolio site showcasing author's VR/AI/full-stack work. TypeScript + React with Three.js 3D scenes, theme switching, and motion animations. Minimal stars/forks but demonstrates multi-domain technical breadth (7 projects across VR, AI/ML, full-stack).
mhosigiri /
skyline-movies
Educational Jupyter Notebook exploring skyline algorithms and data mining concepts. Single notebook (~363 KB), minimal commit history (3 of last 30), no tests/CI/license, created and pushed same day.
06 · Timeline
- Aug 19, 2021Joined GitHub
- Mar 29, 2025Created gemini-alert-app — A crisis de-escalation web application that uses Google's Gemini AI to provide real-time guidance for managing stressful situations
- Dec 26, 2025Created n8n_local_cluster — This is a copy of the original n8n repo. I use this cluster to help extract viral short form content from long form videos.
- Feb 14, 2026Created Google_buildWithAI — This project helps understand the dynamics of using google cloud and AI Agents to understand Graph RAG, AI Agent Orchestration, and build Event Driven Architecture with Google ADK,
- Feb 25, 2026Created skyline-movies — This is a data inference for learning how skyline and Prominent Streak works in Data Mining and Machine Learning.
- Mar 28, 2026Created ChhayaAI — A native iOS emergency response app with real-time location sharing, AI-powered assistance, and a multi-agent backend. Built for situations where seconds matter.
- Apr 10, 2026Created mhosigiri — This is the Official portfolio website for Anish KC (Arnis). Arnis is a VR and LLM researcher as well as a computer science student at UT Arlington. Arnis has been building softwar
- Apr 11, 2026Created HerdSignal — Separate Elephant audio with its filtered spectrogram and annotations detailing the timestamps for when the elephant rumbling happens.
- May 11, 2026Most recent push to n8n_local_cluster
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.