Skip to content
View VibhorGautam's full-sized avatar
14 followers · 15 following

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair Extraordinairex2

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair Extraordinairex2

Organizations

@NeuralAi-DTU

Block or report VibhorGautam

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.

Content in all repositories owned by your account will be closed.
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
VibhorGautam/README.md

Vibhor Gautam

GSoC 2026 @ C2SI | AI agent security | TensorFlow contributor

I work on security, evaluation, and reliability for LLM agents.

During Google Summer of Code 2026, I built the adversarial validation path for ACF-SDK, an open source security firewall for LLM agents. The integration harness exercises the real signed IPC and sidecar pipeline across all 4 agent hooks. It also led to merged fixes for embedded Base64 attacks, fail-closed memory policy, destination allowlists, signal attribution, and thread-safe scanning.

Projects

  • injectlint: pipe untrusted text in and inspect prompt-injection signals before it reaches an agent
  • memory-sota: long-term memory engine with a reproducible LongMemEval-S evaluation pipeline
  • llm-serving-bench: closed-loop serving benchmark for TTFT, inter-token latency, and throughput across concurrency

I also have merged contributions in TensorFlow, OpenClaw, OWASP Juice Shop, OpenSRE, and Hexo.

Contact

Pinned Loading

  1. injectlint injectlint Public

    Pipe untrusted text in and watch prompt injections light up red in your terminal. A best-effort detector, not a guardrail.

    TypeScript 3

  2. memory-sota memory-sota Public

    Long-term memory engine for LLMs. 80.2% on LongMemEval-S (full 500, official gpt-4o judge). Reproducible in one command.

    Python 1

  3. acf-sdk acf-sdk Public

    Forked from c2siorg/acf-sdk

    The Cognitive Firewall is a Zero Trust control layer for agentic systems that protects the reasoning control plane of large language model (LLM) agents. It prevents prompt injection, context manipu…

    Go

  4. GSoC GSoC Public

    Forked from c2siorg/GSoC

    SCoRe Lab GSoC Final Reports

  5. opensre opensre Public

    Forked from Tracer-Cloud/opensre

    Build your own AI SRE agents. The open source toolkit for the AI era ✨

    Python

  6. tensorflow tensorflow Public

    Forked from tensorflow/tensorflow

    An Open Source Machine Learning Framework for Everyone

    C++