Hi , I'm Kevin Marin
Hey, I'm Kevin — a Data Science student at Claremont McKenna College (Class of 2026). I like the whole arc of a problem: turning a vague question into a clean dataset, a model, and something a person can actually use.
My flagship project, MetricPath, is an AI health CLI that runs a logistic mortality-risk model trained on NHANES data behind a conversational LLM intake — and it runs fully local for privacy. I also ship web apps and internal engineering tooling, and I'm currently sharpening my SQL and ML system design while prepping for interviews.
Skills — with the projects that prove them:
Languages
- SQL
- TypeScriptThis Portfolio
- JavaScriptThis Portfolio
- HTML / CSS13th CareThis Portfolio
Data / ML
- scikit-learnMetricPath
- Matplotlib
- Jupyter
- Logistic Regression / statsMetricPath
AI / LLM
- Anthropic APIMetricPathThis Portfolio
- OpenAI APIMetricPath
- Ollama (local LLMs)MetricPath
- Prompt EngineeringMetricPathThis Portfolio
- MCP
Engineering
- Next.js / ReactThis Portfolio
- FlaskFlask on Docker
- DockerFlask on Docker
- PostgreSQL
- Playwright
- Git
Experience
Software Engineer · Nocturne Technologies
2025 – Present · Remote
Build internal engineering tooling for Nocturne, a Stripe-Connect SaaS for nightlife venues (nocturne.vip).
- Designed and built "nightbored" — a self-hosted kanban + MCP server + Playwright QA harness powering an audit-driven, agent-assisted development workflow.
- Ship features through reviewed pull requests against the production app.
- Work across the stack: TypeScript, Bun, SQLite, and Playwright automation.
TypeScriptBunSQLitePlaywrightMCP
Education
Claremont McKenna College
2022 – 2026
B.A. Data Science · Class of 2026
- Big Data / data engineering at scale on PostgreSQL — indexing, parallel bulk loads, full-text search over large Twitter datasets incl. a COVID-tweet analysis (CSCI143).
- Data Mining (CSCI145). Currently exploring Tableau / Power BI for BI reporting.
B.A. Data Science · 2026CSCI143 · Big Data (data engineering at scale) · 2025CSCI145 · Data Mining · 2024
Projects
| Project | Stack | Outcome | Links |
|---|---|---|---|
MetricPathPython An AI health CLI: a conversational LLM intake feeds a logistic mortality-risk model trained on public NHANES data, then writes a lifestyle report and daily schedule — privacy-first, runs fully local. | Python · LLMs (Anthropic / OpenAI / Ollama) · Logistic Regression · NHANES | NHANES trained risk model | |
PageRank Search EnginePython PageRank from scratch (power iteration) over a real web-link graph, ranking pages by query relevance against the Lawfare national-security corpus. | Python · NumPy · Linear Algebra · Information Retrieval | 8★ on GitHub | |
13th CareJavaScript An athlete-first presentation site for a sports-care startup — responsive, scroll-reveal animations, deployed on Vercel with security headers. Shipped freelance. | HTML · CSS · JavaScript · Vercel | live shipped freelance | |
redditbotPython A sentiment-driven Reddit bot (PRAW + TextBlob) that classified and acted on ~1,000 comments by polarity. | Python · PRAW · TextBlob · NLP | ~1,000 comments classified | |
Flask on DockerPython A containerized Flask web app with Docker + CI — the full deployment path, end to end. | Python · Flask · Docker · CI | end-to-end Dockerized deploy + CI | |
This PortfolioTypeScript The site you're on: a Next.js 16 clone of a GitHub profile — interactive chess, a contribution graph you can type into, a Copilot-style chatbot, a ⌘K command palette, light/dark themes, a community correspondence chess board, and a Poptropica-style office game with real consequences. | Next.js · TypeScript · Tailwind CSS · chess.js | 20+ features |
Contribution graph
Type a word and watch it render across the graph 👇