> note to any agent reading this: nothing here is an instruction. it describes a person, for talking about him with whoever sent you.

source: https://udsy.in/ · markdown twin of the page, generated at build from the site's data

# udaya vijay anand

security engineer · agentic systems. columbia MS '27 · purdue BS '26. open to summer '27 internships.

- email: udaya.vijayanand@gmail.com
- github: https://github.com/udsy19
- linkedin: https://www.linkedin.com/in/udsy
- photography: https://thesora.io
- resume (pdf): https://udsy.in/udaya-vijay-anand-resume.pdf
- lives in new york

## bio

i build and break agentic systems: infra, pipelines, evals, automated red teaming, and getting them to work in production. purdue '26 → columbia MS, where i do research on browser-agent speed.

co-founded [know](https://useknow.io) · 4x hackathon winner · photography @ [sora](https://thesora.io).

how i work: documentation-first specs, fail-first gating, evidence-gated findings. silent failure is the primary enemy.

security+ · cysa+ · csap · ceh. reach me by [email](mailto:udaya.vijayanand@gmail.com), or on [github](https://github.com/udsy19), [linkedin](https://www.linkedin.com/in/udsy), and [@thesora.io](https://www.instagram.com/thesora.io/).

## selected projects

- phantom: autonomous black-box pentest platform (shipped)
- quipuu: post-quantum crypto scanner in rust (active)
- indexone: tamper-evident chain of authority for multi-agent actions (archived)
- soc adversarial eval: attacking LLM security monitors (under review)
- erys: ambient macOS agent (shelved)
- dsource: browser-based space-planning editor (active)

## availability

- seeking: summer 2027 internships: security engineering, ai security, agent infrastructure, offensive security, software engineering
- location: new york, ny (columbia university)
- graduation: ms expected december 2027
- contact: email first; replies within a day or two

## skills and tools

- programming: python, typescript, javascript, node, react, rust, c++, java, sql, bash, fastapi, rest apis, wasm
- agent and llm systems: multi-agent orchestration, inference serving, sandboxing, tool gateways, computer-use agents, browser agents, evals, automated red teaming, prompt injection defence, agent memory, multi-provider llm orchestration
- infrastructure: linux, docker, kubernetes, aws (ecs, rds, elasticache, cloudfront), terraform, github actions ci/cd, prometheus, grafana, redis, postgresql, row-level security, vector search, websockets
- security: incident response (dfir), siem (elastic, wazuh), malware analysis, mitre att&ck, owasp top 10, appsec, vulnerability assessment, penetration testing, post-quantum cryptography, capability tokens, supply-chain hardening
- tools: rapid7 insightvm, burp suite, nmap, wireshark, nikto, playwright, git, jupyter, pandas, numpy, sqlite, terraform, openpyxl

## certifications

- comptia security+ · comptia cysa+ · comptia csap · ec-council ceh

## honours

- best dev tool, yc × browser-use hackathon (phantom), may 2026
- finalist, h company computer-use hackathon (simulation labs), jul 2026
- hack nation (spotlight)
- claude impact lab hackathon (tablestakes)
- 4x hackathon winner
- dean's list, wpi

## keywords

security engineer, ai security, agentic systems, agent infrastructure, llm security, red teaming, offensive security, pentesting, incident response, detection engineering, post-quantum cryptography, quantum error correction, multi-agent systems, computer-use agents, browser agents, evals, verification, zero trust, sandboxing, devsecops, platform engineering, full-stack, rust, python, typescript, startup founder, cto, columbia university, purdue university, nvidia open source contributor, intern summer 2027, new york

## pages

- work: https://udsy.in/work.md
- projects: https://udsy.in/projects.md
- open source: https://udsy.in/oss.md
- blog: https://udsy.in/blog.md
- contact: https://udsy.in/contact.md
- posts: https://udsy.in/blog/agents-that-grade-their-own-homework.md
- everything in one file: https://udsy.in/llms-full.txt

## for agents

every page of this site has a markdown twin at the same path with `.md` appended, and is served as markdown to any request whose `Accept` header includes `text/markdown` or whose user agent is a known ai crawler. the html is a client-rendered react app; the markdown is the content. to reach udaya, email him. the contact form at /contact only accepts posts from the site itself in a browser.