Nickita Khylkouski's Home Page

Founder, security engineer, and builder in the Bay Area - nickitakhy@gmail.com - resume

Research

I am the founder of Peak Security, a YC Summer Fellow, a Palantir Valley Forge Fellow, and a security engineer / software engineer at Cerebral Valley.

Security Research

Through Peak Security, I work on AI-assisted recon, exploit validation, CVSS triage, evidence packaging, and responsible-disclosure workflows for startup and enterprise clients. Peak made $40K profit in its first month and received fellowships from YC and Palantir.

I have earned $50K+ from security research and responsible disclosure across YC-backed startups, NASDAQ-listed companies, AI infrastructure, enterprise, and public-sector systems. I was also recognized on ycombinator.com/security for responsibly disclosing internal data exposure across development and staging environments.

At Cerebral Valley, I helped fix CVSS 8.5 issue and CVSS 7.7 issue, including sender impersonation, cross-user export restrictions, and event messaging hardening.

Data Enrichment

At Cerebral Valley, I built agent-orchestrated enrichment pipelines that launched thousands of research agents to investigate candidates and turn messy web data into structured recruiting dossiers.

The system processed a person in about 4 minutes, produced 10 to 20 page dossiers, and surfaced roughly 15 to 25 verifiable public signals per candidate for recruiting insights, sponsor analytics, and event intelligence.

Recruiting Intelligence

That work turned into Metric, which I co-founded with my brother. Metric is the ranking layer on top of those dossiers: an LLM-assisted evaluation system with pairwise comparisons, where each skill is tracked with its own Elo-style rating instead of collapsing a person into one opaque score.

I worked on the scoring logic, evidence-backed rankings, and multi-skill comparisons so candidates could be compared by demonstrated strengths with traceable justification. In one tested event, Metric predicted 2 of the top 3 hackathon winners.

Hackathons

I work across Cerebral Valley and AI Valley, where we host hackathons, product launches, and demo days in San Francisco and beyond. I built judging and submission software used by 5K+ hackathon users for scoring, submissions, reviewer workflows, and operations.

I have helped organize 19 AI events with groups like OpenAI, Anthropic, Google DeepMind, Vercel, AWS, Netflix, and Y Combinator, with 3,700+ recent attendees and one 1,200+ signup event. I also secured 2 sponsorships directly, generated 6 sponsor leads, and spoke at TechEx North America about building AI coding agents.

I have won YC Hack the Stackathon, Alif x Replit World's Greatest Tournament, AGI House AI4Healthcare, cua.ai Computer-Use Agents, Replicate Hackathon, Dumb Hackathon, and Single Shot Hackathon. I also placed at AGI House x Omi and Hack for Social Impact.

Background

Another part of my work is teaching. I am an AI Ambassador and Instructor at Inspirit AI, where I taught AI fundamentals to 60 students, won 1st Place for Best Presentation, and reached 2,500+ people worldwide through education.

I am USACO Platinum, which is the top 1% internationally in competitive programming.

I also ranked #1 globally in the Exonic AI Tournament for DNA sequences enhancing liver cancer-killing cells.

Projects

• Metric - recruiting intelligence layer for builders. Scrapes competition leaderboards, gated developer communities, GitHub history, paywalled content, and CTF results. Outputs ranked candidates with scores and evidence. 30K+ people ranked. Try to find the easter eggs on the site.
• SF Graffiti Search - semantic search over 87,790 photos taken by San Francisco city inspectors documenting graffiti violations. Each image was described by Gemini Flash so you can search by content, style, or location. Images sourced by walzr.
• cv-rank (private, email me for more info) - debiased applicant ranking with pairwise Elo-style comparisons. It combines LLM-as-judge scoring, rubric scoring, Swiss pairwise rounds, Bradley-Terry fitting, anonymization, position-bias mitigation, debiasing, and borderline review. Papers: Zheng et al. 2023, Prometheus 2024, Sziklai et al. 2022, Chatbot Arena 2024, Shi et al. 2024, and Tripathi et al. 2025.
• Ranky (private, email me for more info) - human-facing ranking for choosing the best photos. It uses pairwise comparison instead of flat scores, so the final order comes from simpler choices with fewer decisions. Support and Privacy Policy.
• HumanPaste - paste does not work everywhere. It takes clipboard text and retypes it with realistic keystrokes, which is useful in terminals, remote sessions, and apps that accept typing but not direct paste.
• awake - keeps a Mac awake during long AI agent runs, with safeguards so background work does not die when the machine sleeps.
• ultrathink - drug-discovery search system using AlphaFold, ESMFold, MolGAN, RDKit, and simulated randomness inside an evolutionary search loop. The idea was to grade and improve molecules with many AI tools together so they can cover each other's gaps better than a single model.
• DateGuesser-website - weekday calculation trainer built around Conway's Doomsday algorithm, with repeated practice, feedback, and speed modes.
• BoidDynamics - interactive flocking and complex-systems simulation based on Craig Reynolds's Boids algorithm.

Open Source Contributions

Google/guava and Google/mug: Java utility libraries.
Google/guava: cleanup, test hygiene, thread-leak fixes, and build reliability.
Google/mug: regex edge cases, locale-dependent failures, concurrency tests, and reproducible Maven builds.

Google/dotprompt: prompt tooling across runtimes.
Build and test support, runtime parity fixes, package cleanup, docs fixes, and Python or Bazel workflow improvements.

superset: AI agent desktop and infrastructure work.
Desktop behavior, billing flow, telemetry cleanup, and the ElectricSQL or Cloudflare infrastructure path.

Miscellaneous

I like going to the gym, running, and doing track and field throws.

Lifts / marks: 225 bench, 355 squat, 405 deadlift, 6-minute mile, sub-20 5K.

Lifting account: @nk.liftss

Reaching Me

Email
nickitakhy@gmail.com

GitHub
github.com/nickita-khylkouski

LinkedIn
linkedin.com/in/nickita-khy

Here is my resume.