About Me

Hi, I'm Jonah Rothman. Studying Computer Science (BA/MS) with a minor in Economics at Boston University. I specialize in web development and AI, with hands-on experience in QA automation, full-stack development, and CI/CD pipelines. I've completed two paid technical internships at PanelClaw and ClearWatt, and built production applications like The Task App, The Study Buddy, and InternAtlas. Passionate about creating scalable software solutions and exploring new technologies.

Programming
Java Python C/C++ JavaScript TypeScript Kotlin OCaml
Web & App
React Next.js Express.js Node.js HTML/CSS TailwindCSS Jetpack Compose
Tools
GitHub Actions MongoDB Atlas Azure Firebase Supabase PostgreSQL Vercel
Testing & QA
Playwright Cypress Selenium Jest Maestro Azure DevOps CI/CD
Major League Baseball (MLB)
Jun 2026 – Aug 2026
Software Engineer Intern, Test Engineering
New York, NY

Building Playwright test automation for MLB's ticketing platform serving 70M+ annual fans across all 30 ballparks.

PanelClaw
May 2025 – Aug 2025
Junior Front-End Developer Intern
North Andover, MA

Led Cypress QA automation, cutting regression test time from a full day to about an hour.

ClearWatt
Feb 2025 – Apr 2025
Software Development Intern
London, UK

Built the first automated testing suite with Maestro and wired it into Azure DevOps pipelines.

First Cambridge Realty Corporation
May 2022 – Aug 2022
Information Technology Intern
Cambridge, MA

Installed a new office network, refreshed hardware, and lifted throughput ~50% with optimized fiber configurations.

Boston University
2022 – Dec 2026
MS & BA Computer Science · Minor in Economics
GPA 3.75 Dean's List every term
Thayer Academy
2018 – 2022
High School Diploma
Braintree, MA

Technical Internships

Deployed Apps

All Projects

Class
Jan – May 2026

Iroko Connector — BU Spark!

AI-powered schema matching engine that maps third-party SaaS data into a unified enterprise risk platform — no training data or labeled mappings required.
  • Built ReMatch, a 3-stage LLM pipeline using text-embedding-3-large and gpt-4o-mini to auto-map third-party schemas to a unified ontology with confidence-scored rankings
  • Developed a Next.js + TypeScript frontend with React Flow for interactive schema visualization and live SSE-streamed mapping results
  • Partnered with Iroko Technologies across 4 agile sprints to deliver a production-ready connector for the OneScor enterprise risk platform
Collaborators: Sean Tomany, Abidul Islam, Vanshika Chaddha, Vincent Lin
Class: CS/DS 519 — Software Engineering (BU Spark!) — Spring 2026
Class
Jan – May 2026

Robust Cross-View 3D Pose Tracking

Classical upgrades to multi-camera 3D human pose estimation that cut joint error 11× under occlusion — no neural networks in the evaluated pipeline.
  • Added IRLS robust triangulation, uncertainty-aware Mahalanobis affinity, and a per-joint Kalman filter — all classical, neural-network-free upgrades
  • Cut joint error under 20% occlusion from 219 mm → 20 mm (11×) and flipped tracking score (MOTA) from −0.76 → +0.87
  • Built a stress-test harness across four corruption types and multi-seed evaluation on two benchmarks; also shipped a real-time MediaPipe exercise feedback demo
Collaborators: Sean Tomany, Jigar Kanakhara, Bhavya Bavishi, Harsha Basavaraj Beth
Class: CS 585 — Image and Video Computing — Spring 2026
Class
Jan – May 2026

Predicting Movie Ratings from YouTube Trailer Comments

Can pre-release YouTube comments predict a film's eventual Letterboxd rating? A 5-class NLP pipeline across 173 movies and 236K comments.
  • Scraped TMDB, YouTube Data API, and Letterboxd to build a 173-movie, 236K-comment dataset with strict pre-release filtering
  • Benchmarked six models; TF-IDF + LogReg won at 0.423 accuracy and 0.401 macro F1, beating frozen and fine-tuned DistilBERT variants
  • Engineered 25 features (VADER sentiment, hype/spam heuristics, credibility weighting) and confirmed via temporal ablation that late-window comments carry the strongest signal
Collaborator: Sean Tomany
Class: CS 505 — Natural Language Processing — Spring 2026
Top Personal
January 2026

The Task App

AI-powered task manager combining natural language reminders, Eisenhower Matrix prioritization, and calendar/iCal integration.
  • Built a Next.js task management app with Firebase Authentication and Firestore for real-time task storage and sync across devices
  • Integrated Azure Foundry GPT-5 for AI-powered natural language parsing—type "remind me Friday" and the app schedules with priority and group
  • Implemented drag-and-drop Eisenhower Matrix for instant task prioritization across Do First, Schedule, Delegate, and Eliminate quadrants
  • Built calendar view with iCal feed subscription for seamless integration with external calendar apps
Top Personal
Dec 2025 – Jan 2026

InternAtlas

Search thousands of tech company job boards in one place with live filters and real-time stats.
  • Built a production job board that crawls 2,500+ company ATS boards and normalizes 600,000+ listings into PostgreSQL with full-text search and filters
  • Automated large-scale crawling via GitHub Actions on a 12-hour cadence, feeding Vercel-hosted UI backed by DigitalOcean PostgreSQL
  • Shipped live experience with split-pane UI, advanced filtering, application tracking, and Google OAuth/Firebase sync
Class
December 2025

MealMap

Plan meals, discover recipes, auto-generate shopping lists, log via barcode/planner/custom, and track calories on a dashboard.
  • Built a Jetpack Compose Android app for smart meal planning, nutrition tracking, and shopping list generation using Material 3 design
  • Integrated TheMealDB and OpenFoodFacts APIs with Retrofit, Room, and StateFlow to power recipe discovery, barcode scanning, and persistent weekly meal plans
  • Delivered a real-time nutrition dashboard, onboarding flow, and robust error handling across camera permissions, offline states, and data entry
Collaborator: Abidul Islam
Class: CS501 Mobile Application Development (Fall 2024)
Top Personal
November 2025

The Study Buddy

Upload notes, generate AI flashcards, play study games, and chat with a course-aware assistant.
  • Deployed an AI study platform using React, Vite, TypeScript, and Express backend, running across Azure app services, DigitalOcean, and Google Firebase supporting 50+ users
  • Developed a scalable Express.js backend using Azure Blob Storage for PDF handling and Azure OpenAI to generate flashcards and context-aware subject-specific chat responses
  • Designed a MongoDB Atlas schema with 7 collections, compound indexes, and repository pattern architecture supporting 100+ notes, 50+ flashcard sets, and 200+ chat messages
  • Implemented CI/CD with GitHub Actions for automated builds and deployments
Collaborator: Sean Tomany
Class
November 2025

NBA Team Outcome Prediction

Predicts team scores and winners with a linear regression pipeline on nba_api data, halving error and reaching 0.818 accuracy.
  • Modeled NBA team scores and winners using nba_api game logs with a reproducible Makefile-driven pipeline
  • Engineered features for rest days, home/away context, and opponent stats to train and benchmark regression and classification models
  • Achieved strong accuracy and visualized trends with notebooks that output cleaned datasets, predictions, and validation plots
Collaborators: Alim Ackura, Ash Payal, Justin Liu, Shawn Xiang, Jonah Rothman
Class: CS506 Data Science and Applications (Fall 2025)
Personal
August 2025

VibeScape

Interactive song map to compare playlists with friends, explore clusters, and generate AI-curated blends and top picks.
  • Built a Next.js and React powered interactive Spotify song map that visualizes a user's library as clusters of songs and playlists with the ability to compare music with friends and create custom blended playlists
  • Created an AI-powered playlist chooser via prompt and designed a unique ranking model for Top Songs
  • Harnessed SOTA AI tools like OpenAI Codex CLI to accelerate development time and enhance the UI/UX
Class
Sep 2024 – Dec 2024

Java Terminal Game Engine

Play everything from Tic-Tac-Toe to turn-based RPGs in a scalable, pattern-driven terminal engine.
  • Developed a terminal-based game engine ranging from simple games to complex turn-based gameplay, including DnD-style mechanics
  • Implemented 50+ Java classes to build a scalable, object-oriented architecture
  • Utilized Strategy, Factory, and Singleton design patterns to enhance flexibility and reduce code duplication
Class: CS411 Software Principles (Fall 2024)
Show More Projects
View All on GitHub