01 · Roasts
Research Lab With No Publications
totalStars=80 spread across 68 repos, totalForks=3, soloPct=100%. You've built a distributed systems research institute where you're the only faculty, the only student, and apparently the only reader. spark-resume has 8 modules and 0 stars.
The 24-Hour Architecture Sprint
resumable-driver: 144MB codebase, ARCHITECTURE.md, STATUS.md, design.md — and 30 commits in a single 24-hour window. Either you're a machine or you pasted a very large file and called it shipping.
445 Issues, 0 Collaborators
totalIssuesYear=445 with soloPct=100%. That's not community engagement — that's a very elaborate to-do list you've been keeping with yourself in public.
58% Graveyard Rate
staleRepoRatio=0.58 means more than half your 68 repos haven't been touched in 2+ years. key-value-db has a creation timestamp and nothing else. The GitHub tour buses are stopping at ruins.
All Depth, No Breadth of Audience
Three generations of Spark resumable execution design (PoC → spark-resume → resumable-driver) with documented architectural lessons, conformance suites, and workspace pinning — and literally zero people watching. The iceberg is there; the ship never came.
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% weight60C
- Quality20% weight72B
- Depth15% weight65C
- Breadth10% weight55D
- Community10% weight50D
03 · Stats
365-day commit heatmap
67 active days
Language distribution
- Scala48%
- Java34%
- Python11%
- Jupyter Notebook3%
- HiveQL1%
- R1%
- Other2%
04 · Numbers
Owned repos
non-fork
55
Commits
last 12 months
253
Followers
31
Joined GitHub
Feb 2020
05 · Top repos
unikdahal /
spark-resume
Scala library for Apache Spark stage/shuffle resumption: pluggable backend-agnostic SPI with multi-module architecture (api, core, spark-3.5, celeborn, redis, fs, integration), comprehensive docs, 304KB, proven via real cross-process testing with shared conformance suites.
unikdahal /
resume-poc
Scala PoC for resumable Spark drivers with Celeborn adoption: typed, well-documented multi-file design demonstrating L1/L2 shuffle recovery against a real local cluster, with structured validation suite (7 test modes), but narrowly scoped experimental project.
unikdahal /
spark-resumable-workspace
A sophisticated multi-generation research workspace documenting distributed Spark resumable execution: comprehensive architectural documentation across 3+ upstream forks (Spark, Celeborn, Iceberg) with 102+ passing Celeborn tests and detailed protocol specifications, but primarily a control plane and documentation repo
unikdahal /
resumable-driver
Scala-based resumable Spark framework with comprehensive documentation (design.md, ARCHITECTURE.md, STATUS.md), CI/tests enabled, but zero adoption signals (0 stars, 0 forks, private workspace focus).
unikdahal /
resume-poc-e2e
Scala proof-of-concept demonstrating Celeborn-backed shuffle-stage adoption in AQE-enabled Spark SQL with driver restart resilience. Minimal external adoption signals; focused experimental verification of specific design gaps across multiple projects.
unikdahal /
basalt
Early-stage query engine scaffolding in Rust with solid architecture documentation and typed code. Lexer, parser, and expression evaluator are foundation-ready, but shipping is incomplete—no data I/O, no distributed execution, no tests yet deployed.
unikdahal /
unikdahal
Personal portfolio README with no code or structure. Describes work experience and projects hosted elsewhere, but this repo itself contains only a README with links to external projects (redis-java, sutine).
unikdahal /
key-value-db
Empty scaffold with no commits, no files, and no documentation. Created and immediately abandoned with zero substance.
06 · Timeline
- Feb 24, 2020Joined GitHub
- Jan 9, 2022Created unikdahal
- Jul 20, 2026Created basalt
- Aug 1, 2026Created key-value-db
- Aug 11, 2026Created resume-poc — Resumable Spark drivers PoC: L1/L2 shuffle adoption against a real local Celeborn cluster
- Aug 16, 2026Created resume-poc-e2e — Combined AQE + SQL + Celeborn end-to-end shuffle-stage adoption demo (resumable Spark driver PoC)
- Aug 18, 2026Created spark-resume — A checkpoint/resume layer for Apache Spark: safely reattach to already-committed shuffle output across process boundaries instead of recomputing.
- Aug 23, 2026Created spark-resumable-workspace
- Aug 26, 2026Created resumable-driver — Resumable Spark applications: shuffle-recovery contracts, Spark/Celeborn/Iceberg fork patches, tasksboard (private workspace, multi-agent)
- Aug 27, 2026Most recent push to resumable-driver
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.