01 · Roasts
The Notebook Monoculture
99% Jupyter Notebook. You have 130 repos and somehow managed to make them all basically the same file type. Your 'breadth' is less a portfolio and more a very long reading list.
Burst-and-Ghost Speedrunner
HLD-Zero-to-Hero: 2 days. LangChain-DeepAgents-Playbook: 1 day. Graph-Zero-to-Hero: 4 hours. AI-from-scratch-to-scale: 4 hours. You're not building repos, you're naming them and moving on.
343 PRs, 0 External Friends
343 PRs this year with soloPct = 100%. That's not community contribution — that's an elaborate way to merge branches into yourself. Two issues opened on other people's code. Two.
The Clearpath-DSA Memorial
Clearpath-DSA: created at 12:49:43Z, last pushed at 12:49:44Z. One second of effort. Zero kilobytes of code. A README with only the repo name. Somehow still has 1 star. From whom?
CI/License Desert
12 repos analyzed. 0 have CI. Fewer than 3 have tests. Zero have a license. You're an ML Researcher at Yale who hasn't discovered that other humans might want to use your code legally.
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% weight80A
- Quality20% weight62C
- Depth15% weight58D
- Breadth10% weight40D
- Community10% weight55D
03 · Stats
365-day commit heatmap
278 active days
Language distribution
- Jupyter Notebook99%
- Python1%
- MDX0%
- TypeScript0%
- CSS0%
- HTML0%
04 · Numbers
Owned repos
non-fork
98
Commits
last 12 months
4,510
Followers
21
Joined GitHub
Jan 2022
05 · Top repos
sdivyanshu90 /
Agentic-AI-Zero-to-Hero
Production-style AI agent curriculum with 25 structured modules pairing markdown lessons with typed Python implementations. Well-organized educational project with working code across 5 phases, but nascent (1 day old, 1 star) and no production evidence outside course context.
sdivyanshu90 /
ThinkAct
A well-architected ReAct agent framework with explicit state machine, strict JSON protocol, and comprehensive safety measures. Clean typed Python with tests and structured layout, but brand-new (created 2026-04-29), minimal adoption, and thin real-world validation.
sdivyanshu90 /
llm-orchestrator
Typed Python LLM orchestrator implementing four-stage CoT reasoning pipeline with revision loops, comprehensive docs, and test suite. Very recent creation, minimal adoption signals but structured as a complete working project.
sdivyanshu90 /
LLD-Zero-to-Hero
Early-stage Python LLD interview prep curriculum with typed, structured solutions (hard: API rate limiter, workflow engine, message broker; easy: LRU, vending machine, logger). 6 commits in 1 day, no tests/CI, but coherent architecture and intentional pedagogical design.
sdivyanshu90 /
mcp-college-counselor
End-to-end agentic system for university admissions counseling combining web scraping (Playwright), SQLite storage, MCP server, and LLM-driven recommendations. Well-structured with typed Python, comprehensive README, and test coverage, but very fresh (2 days old) with minimal adoption.
sdivyanshu90 /
Vector-Database-from-Scratch
From-scratch HNSW vector database implementation with rigorous mathematical documentation, NumPy-based L2/cosine distance kernels, and multi-layer graph traversal. Clean typed Python, no tests or CI. Solo commit within hours on 2026-04-25.
sdivyanshu90 /
Graph-Zero-to-Hero
Educational graph algorithms reference with comprehensive phase-based learning path. Python implementations paired with ASCII-heavy markdown explanations. Very new repo (created Apr 26, 2026, last push same day), no tests/CI, no typing, minimal adoption signals.
sdivyanshu90 /
LangChain-DeepAgents-Playbook
Educational LangChain/LangGraph curriculum project with structured modules and multi-level agent examples. Typed Python with README and clear examples, but minimal adoption (1 star), no tests/CI, brand-new (1 day old), no license.
sdivyanshu90 /
dont-buy-yet
Early-stage React financial literacy web app targeting Indian stock market tips. Conversational Hinglish UI with 7 interactive features (tip decoder, market replay, onboarding). Created 1 day ago with minimal commits and no tests/CI. Solid frontend craft but too new to assess impact.
sdivyanshu90 /
HLD-Zero-to-Hero
Educational curriculum for system design covering 12 modules and 20 design problems with structured learning path. Fresh repo (2 days old) with 411 KB size, no tests or CI, minimal adoption (1 star).
sdivyanshu90 /
AI-from-scratch-to-scale
Educational AI curriculum with structured 7-phase learning roadmap and markdown-based lessons. Just launched (1 star, same-day commits), no tests/CI/license, but coherent documentation and ambitious scope justify minimal quality floor.
sdivyanshu90 /
Clearpath-DSA
Minimal repo with only a bare README stub ("Clearpath-DSA"), created and pushed within same minute, no code files, no tests, no CI, no license—an empty scaffold.
06 · Timeline
- Jan 3, 2022Joined GitHub
- Apr 21, 2026Created dont-buy-yet
- Apr 22, 2026Created Clearpath-DSA
- Apr 23, 2026Created AI-from-scratch-to-scale — From-scratch to production AI: practical notes, implementations, and scalable system design.
- Apr 24, 2026Created Agentic-AI-Zero-to-Hero
- Apr 25, 2026Created Vector-Database-from-Scratch
- Apr 26, 2026Created Graph-Zero-to-Hero
- Apr 26, 2026Created LLD-Zero-to-Hero
- Apr 27, 2026Created mcp-college-counselor — An end-to-end agentic system that scrapes university admissions pages, normalizes the data into SQLite, serves it through an MCP server, and lets an LLM call tools to generate a pe
- Apr 28, 2026Created llm-orchestrator
- Apr 29, 2026Created ThinkAct
- Apr 30, 2026Created LangChain-DeepAgents-Playbook
- May 2, 2026Created HLD-Zero-to-Hero
- May 3, 2026Most recent push to HLD-Zero-to-Hero
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.