01 · Roasts
Portfolio, not pull requests
103 public repos and 130 recent sampled commits make the output real; 3 total stars mean the audience has not arrived yet.
Research lab speedrun
SOR_thesis_code schedules 360 runs and 11 Atari games, yet the test suite is still an imaginary control group.
CI is a rare species
Only the profile repo shows CI; random-experiments, MCTS, dormat_RL, and SOR_thesis_code all ship without it.
Burst-mode builder
MCTS packed 30 sampled commits into Aug 11–14, while Reward-Design-Agent packed 13 into a day. Sustained polish wants a turn.
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% weight65C
- Quality20% weight52D
- Depth15% weight58D
- Breadth10% weight55D
- Community10% weight50D
03 · Stats
365-day commit heatmap
43 active days
Language distribution
- Python66%
- C++11%
- Kotlin10%
- Java6%
- JavaScript4%
- HTML1%
- Other2%
04 · Numbers
Owned repos
non-fork
67
Commits
last 12 months
158
Followers
20
Joined GitHub
Oct 2021
05 · Top repos
Peddinti-Sriram-Bharadwaj /
random-experiments
A zero-star experimental monorepo containing a saddle-point optimization demo plus three substantial Android/on-device ML experiments, with useful per-project READMEs but no visible test, CI, license, or adoption signals.
Peddinti-Sriram-Bharadwaj /
MCTS
A substantial three-track MCTS/AlphaZero Python suite spanning classical games, JAX/Flax self-play, TD-SOR training, and hardware profiling, but it has no observed adoption and was built in a short burst.
Peddinti-Sriram-Bharadwaj /
dormat_RL
A substantial documented PyTorch RL research repository implementing dormant-neuron recycling, multi-head controls, Atari dormancy analysis, and neuron-level circuit experiments, but with no demonstrated adoption, tests, CI, license, or typed Python.
Peddinti-Sriram-Bharadwaj /
SOR_thesis_code
A documented, multi-module Python research codebase implementing SOR-enhanced DQN experiments across LunarLander and Atari, with reproducibility and distributed Optuna tooling but no tests, CI, license, or demonstrated adoption.
Peddinti-Sriram-Bharadwaj /
Peddinti-Sriram-Bharadwaj
A maintained GitHub profile repository with a substantial research-oriented README and one scheduled blog-update workflow, but no shipped source product, tests, license, or meaningful adoption signals.
Peddinti-Sriram-Bharadwaj /
Reward-Design-Agent
A documented, modular Python prototype for multimodal Gemini reward design with PPO, video feedback, memory tracking, benchmarking, and SLURM support, but it has 0 stars, no tests or CI, no license, and only a one-day history.
Peddinti-Sriram-Bharadwaj /
qwen_self_play
A documented research prototype for Qwen self-play PPO on Tic-Tac-Toe, with modular environment, agent, trajectory collection, and TRL trainer code, but no tests, CI, license, or demonstrated adoption.
Peddinti-Sriram-Bharadwaj /
RL_practice
RL_practice is an empty practice scaffold containing only a minimal README and no implementation, tests, CI, license, or typed source.
06 · Timeline
- Oct 12, 2021Joined GitHub
- May 8, 2022Created Peddinti-Sriram-Bharadwaj — Config files for my GitHub profile.
- Jun 2, 2026Created SOR_thesis_code
- Aug 11, 2026Created MCTS
- Aug 24, 2026Created Reward-Design-Agent
- Aug 25, 2026Created RL_practice — practice repo
- Aug 25, 2026Created qwen_self_play
- Aug 26, 2026Created dormat_RL
- Aug 28, 2026Created random-experiments — collection of some small random experiments.
- Sep 7, 2026Most recent push to random-experiments
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.