01 · Roasts
Forked harder than starred
aichessathon-starter has 147 forks to 10 stars: people are taking the starter kit home, but the applause is still buffering.
CI has favorites
beast, Broadway Copilot, Fullhouse Bot, and email-mcp ship checks; imc-algothon and optifluke bring serious trading logic but leave CI at the door.
Commit burst, not a campfire
1,926 yearly commits and multiRepoVolume 248 say you are shipping across the board, even if the heatmap has some dramatic quiet stretches.
Hardware side quest unlocked
Most portfolios stop at web apps; rtl8168-fpga went straight to PCIe BARs, MSI-X, PHY behavior, and Verilator.
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% weight78B
- Quality20% weight77B
- Depth15% weight58D
- Breadth10% weight90S
- Community10% weight80A
03 · Stats
365-day commit heatmap
167 active days
Language distribution
- Python51%
- TypeScript32%
- Kotlin7%
- SystemVerilog4%
- C++2%
- Tcl1%
- Other3%
04 · Numbers
Owned repos
non-fork
28
Commits
last 12 months
1,926
Followers
9
Joined GitHub
Oct 2020
05 · Top repos
advitrocks9 /
beast
Beast is a documented TypeScript monorepo for autonomous AI employees, with bounded agent execution, review-driven procedural memory, Postgres/pgvector storage, SSE streaming, demo isolation, and Trigger.dev orchestration.
advitrocks9 /
broadway-copilot-dashboard
A documented, typed Next.js 15 internal operations dashboard with analytics, user administration, agent trace inspection, Prisma/Postgres data access, Vitest coverage, and Cloud Run deployment automation.
advitrocks9 /
broadway-copilot
Broadway Copilot is a substantial typed WhatsApp AI stylist, live for 250 internal users, with a 13-node agent graph, multimodal analysis, wardrobe search, Redis coordination, and GCP deployment automation.
advitrocks9 /
fullhouse-bot
A substantial, documented 6-max Deep CFR poker bot that finished 15th of 64 in the Fullhouse Hackathon, with reproducible benchmarking, runtime packaging, parity checks, and a production-oriented sandbox design.
advitrocks9 /
email-mcp
A documented local MCP mailbox tool with Gmail/Outlook backends, reversible audited mutations, focused tests, and cross-platform CI; adoption remains unproven at 0 stars and 0 forks.
advitrocks9 /
rtl8168-fpga
A documented, technically ambitious SystemVerilog RTL8168H PCIe/NIC emulation fork with detailed BAR, PHY, ERI, MSI-X, and config-space behavior, plus pinned Verilator CI and simulation artifacts, but only 1 star and no demonstrated external adoption.
advitrocks9 /
aichessathon-starter
A documented AI Chessathon starter with a realistic chess protocol harness, sandboxing, arena evaluation, packaging smoke tests, and CI; the root agent remains an intentionally random baseline.
advitrocks9 /
imc-algothon
A documented 15-module Python trading bot with multi-source pricing, six strategy families, async ETF-first execution, inventory reconciliation, and risk controls; credible competition output but no tests, CI, license, or typed annotations.
advitrocks9 /
agent
A documented, test-backed personal FastAPI/Telegram agent with Postgres, MCP integrations, scheduled learning, and approval-gated messaging; technically substantial but currently has 0 stars and no external adoption evidence.
advitrocks9 /
comp-qm
A well-documented, technically ambitious Python scientific-computing rebuild implementing finite-difference and Chebyshev eigensolvers, with extensive validation against analytic, published, scaling, and convergence benchmarks, but no visible adoption or automated CI/tests.
advitrocks9 /
optifluke
A documented, nontrivial Optibook market-making bot with rate limiting, multi-strategy quoting, Black-Scholes pricing, arbitrage, and delta hedging; its reported 9th/30+ academy result shows competition performance but not broad adoption.
advitrocks9 /
advitrocks9
A minimal GitHub profile repository containing only README.md with a short biography and contact links; it has no implementation, tests, CI, license, or adoption signals.
06 · Timeline
- Oct 8, 2020Joined GitHub
- Feb 24, 2021Created advitrocks9
- Aug 12, 2025Created broadway-copilot — WhatsApp AI stylist. Custom StateGraph agent, dual-LLM routing, pgvector wardrobe search, GCP
- Sep 14, 2025Created broadway-copilot-dashboard — Internal ops dashboard for Broadway Copilot. Analytics, cost tracking, agent trace debugger. Next.js 15, shadcn/ui, Prisma.
- Mar 25, 2026Created imc-algothon — Multi-strategy algorithmic trading: tidal-harmonic signals, ETF arbitrage, Monte Carlo pricing. 2nd novice / 7th of 100+, IMC Algothon 2026.
- Mar 25, 2026Created rtl8168-fpga — RTL8168H NIC emulation firmware for the EnigmaX1 (Artix-7 XC7A75T). PCIe Gen2 x1, full config space, BAR register emulation, MSI-X, PHY/MDIO, timing jitter.
- May 14, 2026Created fullhouse-bot — 6-max NLHE poker bot on Deep CFR. MCCFR-trained, pure-NumPy runtime, 2s/768MB sandbox. 12th of 500+, 15th of 64 in the final, Fullhouse Hackathon 2026 (Quadrature Capital).
- Jun 9, 2026Created email-mcp — Local MCP server for reversible Gmail and Outlook mailbox actions.
- Jun 17, 2026Created optifluke — Optiver Trading Academy 2026 market-making bot for the Optibook simulator. 9th of 30+.
- Jun 29, 2026Created beast — Autonomous AI employees that ship real work, learn your voice, and run in the background
- Jun 29, 2026Created agent — 24/7 personal agent over Telegram: life and career pillars, tiered-autonomy messaging, an MCP tool server, and a self-learning loop. Python, FastAPI, Postgres.
- Aug 3, 2026Created comp-qm — Chebyshev pseudospectral eigensolver for the Schrödinger equation of hydrogen in intense magnetic fields
- Aug 28, 2026Created aichessathon-starter — Fork this to build an AI Chessathon agent.
- Sep 6, 2026Most recent push to aichessathon-starter
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.