Upi Shanker

Upamanyu (Upi) Shanker

CS & Finance at Florida State · Software Engineer Intern at FSU ITS · CTO at Campus Sublet

Upamanyu Shanker

About

I'm a dual Computer Science and Finance major at Florida State, minoring in math. Right now I build internal Java apps for FSU IT Services and run engineering at Campus Sublet, a student sublet marketplace with 150+ users. Before that I trained a model at St. Jude that turns plain-language questions into clinical dashboards. On my own time I build things that are fun to take apart: a reinforcement-learning blackjack agent, a stat search over every NBA game, and a movie search that understands vibes.

Education

Florida State University — Dual Degree B.S. Computer Science and Finance, Minor in Mathematics (Expected May 2028)

Honors & Awards

  • National Merit Finalist
  • AP Scholar with Distinction
  • Vires Scholarship
  • Dean's List

Relevant Coursework

Core Computer Science

  • Data Structures, Algorithms & Programming I COP3330
  • Data Structures II COP4530
  • Theory of Computation COT4420
  • Computer Organization I CDA3100
  • Database Systems COP4710
  • Advanced Programming with Java COP3252
  • Pattern Recognition CIS4930
  • Ethics in Computer Science CIS3250
  • Introduction to Programming COP3363
  • Deep and Reinforcement Learning Fundamentals CAP5619
  • Operating Systems COP4610

Math & Quantitative

  • Discrete Mathematics I & II MAD2104, MAD3105
  • Applied Statistics for Engineers & Scientists STA3032
  • Calculus II MAC2312

Browse this site as a shell — type help to start.

Open full screen →
guest@portfolio:~$

click the canvas · space or ↑ to jump · ↓ to duck

Experience

Florida State University Information Technology Services

Software Engineer Intern · Aug 2025 – Present

  • Built a searchable Java (GWT) directory linking every FSU-owned internet domain to its security contact, giving the IT security team one lookup point during incidents instead of manual searches and slashing response times.
  • Building a check-in and appointment scheduling system for 3 university departments, replacing scattered scheduling with a single shared queue used by front-desk staff and visitors.
  • Partnered with stakeholders across departments to take both apps from requirements to deployment, cutting manual processing time by 57%.

Campus Sublet

Chief Technology Officer · Mar 2026 – Present

  • Launched a mobile sublet marketplace for FSU students, scaling to 150+ users on React Native/Expo + Supabase.
  • Built real-time messaging, push notifications, and email-verified auth across iOS and Android.
  • Designed the end-to-end data architecture: 5+ tables with row-level security for listings, chats, and photos.

St. Jude Children's Research Hospital

AI Engineer Intern · Jun 2025 – Jul 2026

  • Delivered clinical reports built from Epic patient records for clinical teams, standardizing hundreds of healthcare metric definitions so reports drew on consistent, validated numbers.
  • Trained a custom AI model that turns natural-language prompts into dashboards for clinical teams, cutting reporting time by 43%.
  • Automated reporting over patient and operational data in Python and SQL, replacing manual spreadsheet pulls.

Florida State University

Research Assistant · Oct 2024 – Present

  • Developed a Python script to categorize duplicate titles in a SQLite database, reducing manual data entry by 75% and automating CSV report generation.
  • Created a SQLite database using NLP and TF-IDF vectorization to automatically analyze and score academic papers, and deployed a Flask web application that queries the database.
Earlier experience (+6)

Johnson & Johnson

Technology Awareness Program · Jan 2026 – Mar 2026

  • Built MedConnect, a telehealth-access platform that addresses two problems for rural patients: finding in-network care and repeating their medical history at every visit.
  • Included insurance-based doctor search, booking, reminders, and a referral system between doctors.

Commerce Hive LLC

Software Engineer Intern · Feb 2025 – Jun 2025

  • Developed modular and accessible front-end components using TypeScript and Next.js, enhancing user workflows.
  • Integrated RESTful API endpoints to fetch and update inventory data, implementing robust error handling.
  • Optimized front-end performance, reducing initial page load time by 52% through dynamic imports, image optimization, and caching strategies.

The DeVoe L. Moore Center

Data Analytics Intern · Jan 2025 – May 2025

  • Created Tableau data visualizations for floridaopengov.org and other Center projects.
  • Collected and cleaned data from a variety of local and state agencies.
  • Conducted market-focused research and supported ongoing transparency projects.

Outlier / Alignerr

AI Trainer · Oct 2024 – May 2025

  • Improved AI performance and accuracy by designing effective prompts and refining LLM responses, ensuring alignment with user needs and project goals.
  • Annotated and labeled video datasets to train and enhance AI model performance, ensuring high-quality data for accurate training and validation.
  • Developed and optimized code implementing machine learning models, conducting rigorous evaluation to identify biases and improve performance.

Voxel Labs, LLC

Front-End Web Developer · Feb 2025 – May 2025

  • Designed and developed a responsive website using HTML, CSS, and JavaScript, ensuring seamless navigation across devices.
  • Integrated interactive elements such as form validation and real-time updates, enhancing functionality and user experience.
  • Deployed and secured the site on AWS: S3 hosting, Route 53 DNS, CloudFront CDN, ACM SSL, and API Gateway + Lambda for backend functionality.

Beats by Dre

Data Analytics Extern · Nov 2024 – Feb 2025

  • Developed and automated data pipelines for gathering, cleaning, and analyzing large consumer datasets using Python.
  • Conducted exploratory and sentiment analysis on customer reviews with pandas and NumPy to uncover actionable insights.
  • Created data-driven presentations and interactive dashboards communicating complex insights to stakeholders.

Leadership

Management Leadership for Tomorrow

Career Prep Fellow · Jan 2026 – Present

  • Selected for a competitive 18-month leadership development program with structured coaching and mentorship.

Asian American Student Union

Programming Coordinator · Jun 2026 – Present

Lambda Phi Epsilon

Webmaster & Events Chair · Jun 2026 – Present

  • Coordinate chapter events and maintain the fraternity website, driving member turnout.

CompNeuroSociety at FSU

Vice President & Webmaster · May 2025 – May 2026

  • Oversaw society operations and strategic planning as Vice President, coordinating meetings and events.
  • Designed, developed, and maintained the organization's website, integrating 3D visualizations and interactivity.
  • Optimized the site for accessibility and mobile compatibility.

Association for Computing Machinery (ACM) at FSU

Webmaster · Aug 2025 – May 2026

  • Led management and continuous improvement of ACM at FSU's WordPress site — content, plugins, and security — for 1000+ student members.
  • Collaborated with the executive board to translate organizational needs into web solutions, increasing event visibility and member engagement.

FSU Association of IT Professionals (AITP)

Outreach Chair · Aug 2025 – May 2026

  • Organized and represented AITP at campus outreach events, driving interest in professional development within IT.
  • Strengthened alumni-student connections by coordinating an alumni interview segment and streamlining event RSVPs.

Projects

Seven I’d be happy to walk through line by line, and the rest below.

Blackjack AI

C++17 blackjack engine with Q-Learning and Monte Carlo agents trained over 2M episodes, playing within ~0.01 units per hand of optimal basic strategy after diagnosing and fixing reward-shaping and learning-rate defects worth ~40% of expected value. Compiles to WebAssembly via Emscripten, so the interactive demo runs entirely in the browser with no backend.

C++Reinforcement LearningMonte CarloWebAssemblyEmscriptenJavaScript

Marquee

Search movies the way you'd describe them to a friend — "feel-good heist", "90s horror", "rating>8.5 war". A hand-written parser applies exact filters in code, while TypeSafe's Jev decision model judges the IMDb top 1,000 against the rest of the query in 11 parallel batches (~0.9 s, ~1¢ per uncached search), tuned on a custom eval harness to 92% precision and 99% recall. Matches rise out of a 60 fps matter-js poster pile onto a ranked shelf, with an accessible grid view and where-to-watch data for 997 of the 1,000 films.

TypeScriptCanvas 2Dmatter-jsTypeSafe (Jev)PythonTMDB APIVitestVercel
Live demo →Source available on request

Statline

Search every NBA player-game in league history by stat range ("20–30 pts, 10–15 ast, 5–10 reb"), with results ranked by closeness to the requested ranges through a dynamically generated SQL expression. SQLite database ingested from stats.nba.com, served read-only behind a Next.js API route and deployed as a Docker image with a persistent volume.

Next.jsTypeScriptSQLitePythonDockerFly.io
Live demo →Source available on request

Cargo Search

Microservice freight-search platform: Spring Boot shipment and carrier services backed by PostgreSQL, a real-time indexing pipeline over Kafka, and an ElasticSearch query layer. Batch analytics and reindexing run on Apache Spark, with every service isolated in its own Docker container.

Spring BootJavaMicroservicesPostgreSQLKafkaElasticSearchApache SparkDocker
Source available on request

Shuttle

Moves files and text between your laptop and phone — scan a QR code once, then push either direction. Browsers upload straight to Cloudflare R2 through presigned URLs and access runs on stateless HMAC-signed tokens, so bytes never touch the server and there is no database at all.

Next.jsTypeScriptCloudflare R2HMAC-SHA256PWAVitestVercel
Live demo →Source available on request

GradeHub

Full-stack grade database with a Next.js frontend and Spring Boot backend over PostgreSQL. Computes GPA and predicts the grades needed to hit a target, with two-factor auth, Google OAuth, and per-student data persistence.

JavaSpring BootPostgreSQLREST APIsNext.jsTypeScriptSupabase

FSU Crime Map

Daily-updated interactive map of FSU police crime-log incidents. A Python pipeline scrapes, deduplicates, and geocodes FSU PD's rolling 60-day log into a growing historical dataset, published as a static Leaflet site with filters, search, a sortable table, and stats. Refreshed automatically by GitHub Actions — no server, no API keys.

PythonpandasgeopyLeafletGitHub Actions
Also built (+6)
  • Clash Royale Stats Tracker · Next.js, Express.js, TypeScript, Tailwind
    React/Next.js frontend with a Node/Express backend that fetches and visualizes player profiles, achievements, badges, and battle logs from the official API.
    Live demo →Source available on request
  • Real Estate MLS · SQL, PHP, MariaDB
    MySQL-backed PHP application to search, view, and query property listings, with a complete relational schema and dynamic pages. Built for a Database Systems course.
  • YT → Tagged MP3 · TypeScript, Python
    Converts YouTube videos into MP3s with proper ID3 tags and cover art.
  • Outlook–Google Calendar Sync · Python, Microsoft Graph API, Google Calendar API, SQLite
    One-way syncs multiple Outlook calendars to Google Calendar, handling recurring events, color coding, and duplicate prevention via a SQLite tracking database.
    Source available on request
  • NLP Paper Scoping Tool · Python, NLP, SQLite, Flask
    TF-IDF and SQLite pipeline with a Flask UI to score and filter academic papers, with automated duplicate detection. Built for FSU research.
    Source available on request
  • FloridaOpenGov Visualizations · Tableau, Data Visualization
    Interactive Tableau visualizations of Florida economic, school, and housing data for transparency projects.
    Source available on request