Skip to content
View VarunGore36's full-sized avatar
🎯
git commit -m "been there, still doing that"
🎯
git commit -m "been there, still doing that"
153 followers · 346 following

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair Extraordinairex2Achievement: YOLO

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair Extraordinairex2Achievement: YOLO

Block or report VarunGore36

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
VarunGore36/README.md
Varun Gore — systems, security, and agent infrastructure

LinkedIn  ·  Email  ·  GitHub  ·  Discord


I build infrastructure that does more than work—it can account for itself. My interests sit at the intersection of systems engineering, agent security, reproducible measurement, and developer tooling.

Live signal

01 / CURRENT FOCUS

Agent systems that fail safely

Constrained tools, least-privilege credentials, controlled network access, and evidence for every consequential action. 		02 / DEFAULT MODE

Evidence before confidence

Benchmarks need context. Claims need boundaries. Unverified results stay visibly unverified. 		03 / BUILDING WITH

Rust · Python · Linux

Systems software, reproducible research, AI evaluation, and tooling people can inspect and trust.

Selected work

Toolcage

LOCAL-FIRST AGENT SECURITY RUNTIME

A Rust runtime that puts policy, credential brokering, controlled egress, and tamper-evident audit logs between an AI agent and everything it touches.

Rust Linux MCP security

72 tests · 404 ns policy-decision p99 · alpha / not independently reviewed

REA

OPEN-SOURCE CONTRIBUTION · AGENTIC REVERSE ENGINEERING

Contributed to a project for reverse engineering software with agents, from observed behavior down to native binaries.

TypeScript agents reverse engineering

Focused on making deep analysis workflows more reliable and usable from the command line.

ECC

OPEN-SOURCE CONTRIBUTION · AGENT HARNESS ENGINEERING

Contributed to an agent-harness optimisation system spanning skills, memory, security, and research-first development workflows.

JavaScript Codex developer tooling

Built for practitioners working across Claude Code, Codex, Cursor, OpenCode, and related environments.

AstralProject

REPRODUCIBLE CRYPTO-MARKET MEASUREMENT

Lossless market-data capture, deterministic replay, and honest cost models for research and simulation.

Rust market data replay

300 / 300 top-of-book checks · live project

ChainLens

ETHEREUM INDEXING, BUILT FOR FAILURE MODES

A concurrent Rust indexer with canonical-chain tracking, reorg handling, crash-safe recovery, and PostgreSQL storage.

Rust Tokio PostgreSQL

232 tests · 150 ns/block decode benchmark · live project

ENAMEL-Extended

LLM CODE-EFFICIENCY EVALUATION

An open reimplementation and extension of the ENAMEL benchmark for LLM-generated code.

Python evaluation LLMs

13 models · 161 problems · 612 tests · live project

More experiments: DepLens — testing whether dependency graphs, code usage, and Git history can predict what a change will break.



A system is only as trustworthy as its failure mode



How I build

Determinism before speed

Same input, same output, every replay. Optimisation is only interesting when the system remains explainable.

Fail closed

If an action cannot be authorised, isolated, or recorded, it should not quietly proceed with weaker guarantees.

Measure the boundary

Every result has a method, environment, and limit. A benchmark without context is marketing, not evidence.

Keep the rough edges visible

Incomplete work, uncertain conclusions, and known failure modes are part of the engineering record.

Technology stack: Rust, Python, TypeScript, SQL, Linux, Tokio, Axum, SQLx, Alloy, vLLM, EvalPlus, PostgreSQL, Docker, GitHub Actions, Prometheus, Grafana, pytest, Ruff, uv, NumPy, and Matplotlib


Find Varun Gore on Discord


Build systems. Leave evidence.

Pinned Loading

  1. ECC ECC Public

    Forked from affaan-m/ECC

    The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

    JavaScript 3

  2. Crukx-dev/CrukxCLI Crukx-dev/CrukxCLI Public

    Rust 1

  3. DepLens DepLens Public

    DepLens is an investigation into dependency-change impact prediction. The objective is not to assume that reliable prediction is possible, but to determine how far it can be pushed using empirical …

    Python 3

  4. rea rea Public

    Forked from morluto/rea

    Reverse engineer anything with agents, from app behavior down to native binaries.

    TypeScript 4

  5. Blue-Intelligence Blue-Intelligence Public

    Forked from Berry-Mappemonde/Blue-Intelligence

    Blue Intelligence — mémoire des sources maritimes

    Python 3

  6. toolcage toolcage Public

    Local-first runtime for securely executing AI agent tools, with capability-based authorization, credential isolation, controlled network egress, and tamper-evident audit logs.

    Rust 5 1