Skip to content
View SuryanshSS1011's full-sized avatar
5 followers · 5 following

Achievements

Achievement: Pair Extraordinairex2Achievement: Pull Sharkx2Achievement: YOLO

Achievements

Achievement: Pair Extraordinairex2Achievement: Pull Sharkx2Achievement: YOLO

Highlights

  • Pro

Block or report SuryanshSS1011

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SuryanshSS1011/README.md

CS @ Penn State.

I research safe and secure deployment of AI/ML systems using reinforcement learning. IEEE AITest 2025, LCTES '26 WIP, CNSM 2026 (submitted).

Currently working on

  • OSS contributions — check my contribution log.
  • Warren — Turn your Wikipedia rabbit hole into a beautiful, shareable map. Next.js 16 + Supabase + Claude Haiku 4.5.

Research

  • match-loss-to-cost — Decision-aware traffic forecasting with asymmetric losses and conformal capacity bands. Submitted to CNSM 2026 (June 1, 2026). Evaluated on Abilene, GÉANT, and CESNET-TimeSeries24 across LSTM, DLinear, iTransformer, PatchTST, and DCRNN.
  • SecureCodeRL — Partial-credit RL for reliable code generation with small language models. LCTES '26 WIP. Partial-credit PPO lifts syntax validity to 63% on APPS+ where binary-reward PPO regresses below the SFT baseline.
  • Performance-Bugs-LLM — Detecting and explaining Java performance bugs with a fine-tuned LLM. 490-bug dataset over 17 Defects4J projects. IEEE AITest 2025.

Built

Activity

GitHub Streak Stats

Footer Wave

Pinned Loading

  1. SecureCodeRL SecureCodeRL Public

    Scheduled Partial-Credit RL for Reliable Code Generation with Small Language Models. LCTES '26 WIP.

    Python

  2. match-loss-to-cost match-loss-to-cost Public

    Decision-aware traffic forecasting for backbone capacity planning. Asymmetric losses and conformal calibration to minimize operator cost rather than RMSE. Evaluated on Abilene, GÉANT, and CESNET-Ti…

    Python

  3. Shift Shift Public

    Shift v0.1.0 was engineered in 12 hours for the GDG @ Penn State Solution Challenge to solve a critical irony: using AI to save the planet while ignoring its energy cost. It balances personalized c…

    TypeScript 1 1

  4. Performance-Bugs-LLM Performance-Bugs-LLM Public

    Detecting and explaining Java performance bugs with a fine-tuned LLM. Dataset of 490 bugs, code, and evaluation. IEEE AITest 2025.

    Python

  5. TruthCast TruthCast Public

    A fully autonomous, multi-agent fact-checking pipeline.

    TypeScript