Sushant Poudel

I find where an autonomous agent's tool boundaries leak, then close them.

sushant.poudel2028@gmail.com · +977 9863635324 · Kathmandu, Nepal (remote, worldwide)
github.com/sushant-me · linkedin.com/in/sushant-poudel2028 · sushantpoudel2028.com.np

Agent & LLM security Memory-safety in C/C++ Evaluation design Edge / offline inference

Email me Check my claims

Every claim on this page is checkable, and checked weekly.
sushant-me/reputation lists each claim I make, names the public source it came from, and re-runs those checks in CI once a week — it turns red the moment one stops holding. Claims that cannot be checked from outside are marked on request rather than mixed in with the verified ones.

git clone https://github.com/sushant-me/reputation && cd reputation && python3 verify_evidence.py

What I can do for you

1 · Agent and LLM security review

Your agent executes actions — calls tools, touches files, moves money, runs commands — and role-based access control authenticates the identity without having any opinion about whether a given action should happen. I review the stack for the failure modes that actually occur: prompt injection through retrieved content, tools that shadow framework built-ins, confirmation gates that fail open, and destructive actions reachable without an independent check.

2 · MCP server audit — before you connect to it

A tool description is not documentation: it is injected into the model's context, so it can carry instructions. I review the declarations your client receives from a server's tools/list — and pin them, so a description that changes after you approve it is reported instead of silently steering the agent.

3 · Memory-safety and fuzz-harness work (C/C++)

Memory bugs in the libraries everything else is built on, plus the harnesses meant to catch them. I filed five memory-safety issues against Google's s2geometry — a NULL dereference on the documented traversal path, an out-of-bounds read, a 16 GiB allocation from a 117-byte input, and a 2.4 GiB allocation reachable from a 28-byte input — each with a standalone reproducer. In libphonenumber I found an OSS-Fuzz harness that was asserting 0.00% line coverage of the code it targeted; the corrected one reaches 93%.

4 · Evaluation design for AI features

"It works" is not a measurement. I design evaluations with controls that can fail: a shuffled-label floor, an ablation that must collapse, a deliberately broken system that must be caught. My current paper ran 600 policy scenarios per condition on a 3.8B model inside a 4.2 GiB budget, and the headline is the result that contradicted my hypothesis — JSON-constrained output without a reasoning field was the least safe configuration of the three, at 46.2% unsafe accepts against 17.2% unconstrained.

How I work

The receipts

whatcheck it
Security patch authored, merged, then generalised into google/go-github — my #4556 refused a release-asset upload whose URL pointed off-host; the maintainer replaced it the next day with the broader #4564, "credentials are sent only to configured origins", which is what master implements today and which carries the commit "Address feedback from sushant-me" #4556 · #4564
Five memory-safety issues in google/s2geometry, with reproducers; two hardening pull requests under review #674, #676#679, PR #682
MCP tool-shadowing fixes open against three of Google's agent frameworks adk-go #1606 · adk-java #1515 · adk-python #7145
Rank #1, HackingHub Q3 2026 global leaderboard — 116 flags, 11,860 XP leaderboard API
A benchmark I wrote found a false positive in my own released scanner — 23 labelled agent tool-boundary cases, precision 0.750 → 1.000, with the failing case named in CI instead of deleted tool-boundary-corpus · the writeup
Two IEEE papers accepted — local policy verification for agents; predictive bandwidth scaling for WebRTC over LEO links (2026 IEEE RTC, Chicago) paper, code, raw outputs
Tools I maintain: agentbound, mcp-nameguard, mcpaudit, policygate, trajectorycheck each has releases, tests and CI
Two years as an AI engineer at Atmos SoftTech — production ML, full-stack delivery, offline-first agent pipelines signed experience letter on request

Availability

Open to contract and freelance engagements now, and to full-time AI security / agent safety roles from graduation. Final-year Computer Engineering at Nepal Engineering College, so I keep one client engagement at a time alongside it.

sushant.poudel2028@gmail.com