UC Berkeley · Research · Engineering · Founder of PoloIQ™

Hi, I'm Jack.

I'm a student in the College of Chemistry at UC Berkeley and a researcher in the Grazioli Computational Chemistry & Biophysics Group at San José State, where I develop computational methods that use molecular simulations and machine learning to study configurational dynamics in biomolecules, molecular self-assembly, and chemical reaction dynamics.

I'm interested in machine learning for chaotic chemical, physical, and biological systems, and in synthetic biology, biochemical engineering, drug discovery, and enzyme and metabolic engineering. That comes from my love of physics and chemistry, and from something personal: a BRCA2 mutation runs in my family. I want to combine my computational skills with hands-on lab techniques to make real change.

I'm also the founder of PoloIQ™, a patent-pending sports analytics company.

Jack Spitzer

About

What I'm doing now

I'm a student in the College of Chemistry at UC Berkeley. I'm also a researcher in the Grazioli Computational Chemistry & Biophysics Group at San José State, mentored by Dr. Gianmarc Grazioli since 2024. I develop computational methods that use molecular simulations and machine learning to study configurational dynamics in biomolecules, molecular self-assembly, and chemical reaction dynamics.

Outside of class and the lab, I'm building PoloIQ™ into a company. It's analytics and coaching software for collegiate water polo, and its core technology is patent pending. I'm also building ReactorMind, an AI study tool for chemical engineering. At Berkeley I'm involved with AIChE, ACS, the ChemE Car team, and the Biofuels Technology Club research team.

What I've done

  • Co-founded Water Polo Stats, a voice-enabled stat tracker used by high schools and clubs across California
  • Presented research at the ACS Western Regional Meeting and NCUR, and published on interpretable ML in medicine in the Curieux Academic Journal
  • Built AI/ML tools for client companies as a developer and consultant at the startup White Lotus

PoloIQ™

PoloIQ logo

Founder · Est. 2026 · Patent pending

Voice-powered stats and real-time coaching insights for water polo

PoloIQ™ is software for coaches, managers, and statisticians. Track every stat and coaching note hands-free by voice, synced to the official game clock, and define your own stats around your team's goals. Then get explainable, data-driven recommendations for lineups, matchups, and plays during the game. PoloIQ™ learns from every game and every outcome, so its recommendations keep getting better the more you use it.

🎙️ "dark 3 excluded by light 2" → Exclusion: Dark #3 vs #2 · Exclusion drawn: Light #2 vs #3 · Q3 4:12
🗣️

Hands-free voice stats

Call out plays the way coaches already talk. PoloIQ™ resolves "dark" and "light" caps to the right team for that game, understands active and passive voice, and logs linked events for both players: who was excluded, who drew it, who assisted, who blocked whom. Say "undo" to take it back.

⏱️

Camera reads the scoreboard

Point a phone at the scoreboard and draw a box around the clock once. PoloIQ™ reads the digits live and re-syncs after every whistle, timeout, and stoppage, so every event lands on the official game clock, not the phone's.

📝

Speak your own stats

Name a player and say anything ("too high in the lane," "great center D") and it becomes a coaching note on that player. Notes a coach keeps repeating turn into measurable trends the engine uses.

🤝

Head-to-head ledger

Every player-vs-player result is tracked across every meeting: goals scored on, exclusions drawn, steals, blocks, sprints won. Each player's timeline shows every event with the period, game clock, score, and the opponent involved.

🧠

Explainable tactics engine

Set up the situation on a drag-and-drop tactics board. PoloIQ™ tests who performs above or below their norm in this exact situation (period, score, man-up or man-down) for statistical significance, scores every matchup, and recommends a substitution, matchup, and play. Each call comes with ranked, plain-English reasons and a confidence level.

📈

Gets smarter every game

One tap answers "Did it work?" PoloIQ™ tracks which plays actually succeed for your team in each situation and re-ranks them, without letting a lucky streak take over. Every signal is recomputed as new games are logged by voice or imported.

📊

Scouting & game plans, automated

Box scores and live-scoring feeds are pulled in daily and merged into one timeline. Before each game, PoloIQ™ builds a plan: the record against the opponent, the last meeting, who has scored against them, and room for staff notes.

🏀

Built for water polo, ready for any sport

The same system carries over to basketball, hockey, soccer, lacrosse, volleyball, football, baseball, and more. Only the vocabulary changes.

U.S. Patent Application No. 64/169,223 · PoloIQ™ is a trademark of Jack Spitzer.

Projects

ReactorMind logo

Creator

ReactorMind

An AI study platform for chemical engineering. It has interactive simulations rebuilt natively with real math, across reactor design, thermodynamics, separations, heat transfer, and more. An AI tutor explains the live numbers on screen, and the platform also generates notes and practice problems.

  • React
  • FastAPI
  • SciPy
  • LLMs
Open ReactorMind →
Water Polo Stats logo

Co-founder & developer · 2024–present

Water Polo Stats

A website and app for voice-enabled stat tracking, with stats, schedules, rosters, and timestamped game logs. It's used by high schools and clubs across California and is the foundation PoloIQ™ is built on.

  • Voice AI
  • Web + mobile
Scraper module diagram

White Lotus · 2025–2026

Social Intelligence Scraper

One of the AI/ML tools I built for White Lotus clients. It's a modular Python pipeline that scrapes, scores, and summarizes LinkedIn, Reddit, and X posts with GPT-4o or local models and sends alerts.

  • Python
  • LLMs
  • Automation
Custom-built drone

Hardware

Custom Drones

Drones designed and built from scratch, from frame and propulsion to electronics and flight software.

  • Hardware
  • Embedded

Research & Publications

PoloIQ™

Patent application · USPTO · Filed Oct 4, 2026 · Patent pending

Systems and Methods for Voice-Driven Sports Event Capture, Camera-Based Game-Clock Synchronization, and Explainable, Self-Improving Lineup and Tactics Recommendation

Inventor: Jack Spitzer · U.S. Application No. 64/169,223

Covers PoloIQ™'s core technology: voice-driven event capture with linked player-vs-player events, camera-based reading of the scoreboard clock, and an explainable recommendation engine that learns from the outcomes of its own advice.

Double pendulum trajectory

Presentation · NCUR 2026

From Double Pendulums to Acetaldehyde Photodissociation: Exploring Chaotic Dynamics with Machine Learning

Vyas, D., Wong, A., Spitzer, J., Dang, H., Huynh, M.H., Grazioli, G.

Uses machine learning on chaotic systems, from the classic double pendulum to bond breaking in acetaldehyde.

ACS poster presentation

Poster · ACS Western Regional Meeting · Oct 2025 · San Jose, CA

Exploring Acetaldehyde Photodissociation Pathways Using Interpretable AI Models

Kim, J.H., Dang, H., Joshi, H., Vyas, D.H., Wong, A., Sridharan, S., Spitzer, J.D., Andreasyan, D., Bojeh, E., Grazioli, G.

Uses interpretable ML on ab initio molecular dynamics trajectories to map how acetaldehyde breaks apart after absorbing light.

Paper abstract

Paper · Curieux Academic Journal · Nov 2025, Issue 56 (Part 2)

Unlocking Transparency: The Role of Non "Black Box" Machine Learning Models in Medicine

Spitzer, J.D.

Argues that diagnostic and treatment AI should be interpretable, and examines transparent models such as SVMs as alternatives to black-box approaches.

Updates

Follow me on X →

Let's connect

I'm always open to conversations, collaborations, and feedback, whether you work in a similar field or you're just curious.