01 · Roasts
The 423 PR Man Who Won't Write a Test
You opened 423 PRs this year — a genuinely staggering number — yet spuma (254 MB codebase), spuma-cases, and figr all ship with zero tests and no CI. You clearly know what CI looks like; ffmt has three integration test files. So why does your GPU turbulence solver not deserve the same courtesy?
Horizontal Sprinter, Vertical Avoider
multiRepoVolume of 163 across 29 repos is impressive breadth, but spuma-cases has 6 commits in 13 days and spuma's entire history is one day. You're great at starting fires; less great at tending them past the first week.
59% C++, 37% C — It's 2026
Your language breakdown is basically 'C with extra steps.' Rust sneaks in at <1% via ffmt, which is genuinely great, but the rest of the portfolio is HPC systems code that would make a 1995 Cray programmer feel right at home. Diversify before the compiler warnings are the most colorful thing in your repos.
Fake-Beamer Has More CI Than Your GPU Solver
A PowerPoint theme replicator has CI that validates slide layout counts and Accent1 hex color invariants (#3331B4, specifically). Your 254 MB CUDA-flavored CFD fork has nothing. The shell script has more discipline than the supercomputer code.
346 Stars on a Template, 5 on Your Actual Tool
academic-website-template is your most-starred repo at 346 stars — a Jekyll theme. ffmt, your most technically impressive work (Rust, PyPI, ecosystem CI, ARCHITECTURE.md), has 5 stars. The internet continues to reward CSS over compilers.
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% weight73B
- Consistency20% weight80A
- Quality20% weight82A
- Depth15% weight68C
- Breadth10% weight55D
- Community10% weight65C
03 · Stats
365-day commit heatmap
312 active days
Language distribution
- C++59%
- C37%
- Shell2%
- Fortran1%
- SCSS0%
- Rust0%
- Other1%
04 · Numbers
Owned repos
non-fork
13
Commits
last 12 months
1,546
Followers
134
Joined GitHub
May 2018
05 · Top repos
sbryngelson /
ffmt
Domain-specific Fortran formatter written in Rust with strong engineering: comprehensive test suite, CI/CD (GitHub Actions), configuration system (TOML), PyPI distribution, and production-ready architecture. Clear niche tool with industry relevance.
sbryngelson /
academic-website-template
Well-structured Jekyll academic website template with 346 stars, used by 200+ academics worldwide. Features CI/CD, responsive design, dark mode, publication management via BibTeX, and comprehensive customization options. Active maintenance with ~30 recent commits.
sbryngelson /
Fake-Beamer
PowerPoint Beamer theme replicator with rigorous shell-based OOXML validation, comprehensive documentation (docs/, ARCHITECTURE.md, STATUS.md), CI/CD, and structured font/design management. Niche indie tool with clear ownership.
sbryngelson /
georgia-tech-letterhead
Specialized LaTeX template for Georgia Tech letterhead with polished branding implementation, well-documented setup, and active maintenance across 4.6 MB of assets and configuration.
sbryngelson /
figr
GPU-accelerated Fortran finite volume solver (IGR mini-app) with comprehensive CI/testing infrastructure, modern build system, but minimal adoption (1 star, 0 forks, <5 days old).
sbryngelson /
georgia-tech-cv
Specialized Georgia Tech CV template in LaTeX with MIT license, Makefile build system, and detailed biblatex integration. 6 stars, minimal adoption. Personal/institutional tool rather than general-audience project.
sbryngelson /
spuma-cases
Minimal supplementary data/config repo for sbryngelson/spuma turbulence closure project; 27MB C++ case files and NN weights with no documentation, tests, or CI, created 13 days ago with sparse commit activity (6 of last 30 days).
sbryngelson /
spuma
GPU-native OpenFOAM fork (SPUMA) by CINECA with fused NN/classical turbulence closures. 254 MB codebase, README present, but created Apr 2026 with only 30 commits on same day; no tests, CI, or license file; untyped C++.
06 · Timeline
- May 29, 2018Joined GitHub
- Mar 23, 2021Created academic-website-template — Jekyll website template for personal academic or research group web pages.
- Dec 5, 2021Created georgia-tech-letterhead — Replicating the GT letterhead in LaTeX
- May 1, 2022Created Fake-Beamer — You like Beamer? You need Powerpoint? Fake Beamer is for you!
- Feb 28, 2023Created georgia-tech-cv — LaTeX CV following Georgia Tech's idiosyncratic RPT (tenure) requirements
- Mar 20, 2026Created ffmt — A fast Fortran formatter, written in Rust
- Apr 2, 2026Created figr — Finite volume IGR mini-app
- Apr 12, 2026Created spuma — GPU-native OpenFOAM fork (CINECA SPUMA) with fused NN/classical turbulence closures for RANS
- Apr 12, 2026Created spuma-cases — Case files, NN weights, and benchmark scripts for sbryngelson/spuma turbulence closures
- May 14, 2026Most recent push to ffmt
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.