01 · Roasts
Test Allergic
2,713 commits this year across 8 scored repos and not a single one has HAS_TESTS=yes. You've written a RISC-V kernel, a TurboQuant SIMD engine, and an io_uring LLM loader — but apparently 'assert' is a forbidden keyword.
Burst Builder Syndrome
tensora: 30 commits in 4 days. turboquant: 18 commits in 3 days. pico-os: 30 commits in ~1 month. You sprint like a caffeinated squirrel, ship something impressive, then vanish. Sustained maintenance is not in the vocabulary.
52% Notebook Hoarder
Over half your GitHub byte-weight is Jupyter Notebooks, yet deep-learning has no README, no CI, no license, and 5 sparse commits. You're paying serious systems-engineering rent in Rust and Zig but sublettig half the profile to unstructured notebooks.
517 PRs, 85% Solo
517 pull requests in a year sounds like open-source royalty — until you realize 85% are you merging your own branches. That's not collaboration, that's a very elaborate way to talk to yourself.
License Roulette
pico-os claims MIT in the README but HAS_LICENSE=no in the repo flags. cpp-kafka has CI but no license. turboquant ships design.md, ARCHITECTURE.md, and STATUS.md but forgot the one file lawyers care about.
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% weight61C
- Consistency20% weight85A
- Quality20% weight67C
- Depth15% weight65C
- Breadth10% weight80A
- Community10% weight50D
03 · Stats
365-day commit heatmap
323 active days
Language distribution
- Jupyter Notebook52%
- Rust14%
- TypeScript11%
- Python4%
- Java3%
- Go3%
- Other13%
04 · Numbers
Owned repos
non-fork
66
Commits
last 12 months
2,713
Followers
96
Joined GitHub
Mar 2021
05 · Top repos
botirk38 /
pico-os
Educational RISC-V OS in Zig with kernel, paging, VirtIO, filesystem, and multi-process scheduling. Well-structured codebase with CI, clean module organization, and working demo—typical portfolio-quality systems project.
botirk38 /
turboquant
A focused Zig implementation of Google's TurboQuant vector compression algorithm with comprehensive API, well-documented architecture, CI/CD coverage, and performance benchmarks. Clean modular design with polar and QJL quantization components, but minimal adoption signal and no license.
botirk38 /
portfolio
Personal portfolio/blog built with Next.js 15, TypeScript, and MDX. Fully documented with CI/CD, type-safe code, and blog infrastructure. Experimental indie project with 1 star, created Sept 2025, 30 commits in 6 months.
botirk38 /
tensora
Early-stage adaptive LLM checkpoint loading framework in Rust with SafeTensors + ServerlessLLM format support, multiple I/O backends (sync, async, io_uring), and Python bindings. Well-typed codebase with structured architecture, but minimal adoption (1 star), no tests, and very recent creation (4 days old).
botirk38 /
sift
Trigram-indexed regex search tool in Rust with typed library + CLI. Structured codebase with docs and CI, but minimal stars/adoption and recent creation (3 weeks old).
botirk38 /
cpp-kafka
Lightweight C++26 Kafka server with protocol parsing, thread-pool client handling, and log storage—experimental single-owner project with working typed code, CI/testing, but minimal adoption and no license.
botirk38 /
botir-skills
One-shot skills collection repo with clear README documenting the skills.sh-compatible format. Only 1 star, 3 recent commits across 2 days, untyped shell-based educational content (rust-performance skill distilled from upstream). Minimal scope and early-stage adoption.
botirk38 /
deep-learning
Jupyter notebook collection on deep learning with minimal documentation, no tests/CI, and sparse commit activity over ~7 weeks. Experimental educational project with 58MB of content but no README or structured output.
06 · Timeline
- Mar 11, 2021Joined GitHub
- Dec 17, 2024Created cpp-kafka — A lightweight, high-performance Kafka server implementation in modern C++23.
- Feb 24, 2025Created pico-os — A minimal RISC-V 32-bit OS written in Zig.
- Sep 23, 2025Created portfolio — My portfolio/blog :)
- Jan 29, 2026Created deep-learning
- Mar 23, 2026Created sift — Indexed regex search over a codebase: build a trigram index once, then query it with a grep-like CLI or the sift-core library.
- Mar 26, 2026Created turboquant — Library for Google's Turboquant Algorithm
- Mar 28, 2026Created botir-skills — My skills repo
- Apr 15, 2026Created tensora — A blazingly fast LLM checkpoint loading framework
- Apr 18, 2026Most recent push to tensora
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.