Sheel Shah

Matchday program · Season 26/27

Sheel
Shah

Computer science student. I build software, machine learning models and things with wires.

The move0/6

Warming up…
Sheel Shah
No. 26, Sheel Shah

Player profile

I like building things people actually use. Lately that has meant a schedule builder for UMD students, a Raspberry Pi that drives an electric car, and a model that correctly predicted the 2026 World Cup. When I'm not at a keyboard I'm usually watching soccer.

Position
Software, ML, hardware
Strong foot
Python
Also plays
TypeScript, Java, C, SQL
Kit bag
React, FastAPI, Dash, scikit-learn, XGBoost, pandas, Raspberry Pi

Team sheet

The starting seven. Pick a player on the field, or read on.

  1. 10

    OrbitTerp 2026

    A schedule builder for University of Maryland students. You pick your courses, block off the times you're busy and choose what matters most, like professor ratings or gaps between classes. It searches the possible schedules and gives back the best ones, ready to export to your calendar.

    React / TypeScript / FastAPI / Qiskit

  2. 9

    Autonomous EV, phase 1 2024–25

    A Raspberry Pi that drives a real electric car, built with a partner at Arlington Career Center for Switch Labs. The Pi sends a controlled voltage through an MCP4725 DAC in place of the gas pedal, runs off the car's 12 V battery and has a touchscreen control panel. It drives forward and backward with nobody in it, won multiple trophies and $10,000 in competitions, and got us on live TV. Braking, steering and computer vision are the next three phases.

    Python / Raspberry Pi / I²C / tkinter / Circuits

  3. 7

    Aegis Orbit 2026

    A hackathon project with friends that predicts when satellites might hit space debris and plans how to move them out of the way. I worked on the data side: pulling CelesTrak data for more than 18,000 tracked objects, cleaning it, and training an XGBoost model that flags the most dangerous ones. Every flag gets checked against the real physics.

    Python / XGBoost / FastAPI / React / CesiumJS

  4. 8

    World Cup Predictor 2026

    Simulates the 2026 World Cup from the group stage to the final, trained on more than 8,000 international matches. I tested a classical XGBoost model against a quantum one on 987 games neither had seen. XGBoost called 57.5% of them right and the quantum model 42.5%.

    Python / XGBoost / scikit-learn / pandas / Qiskit

  5. 6

    Home Board 2026

    The screen on our apartment wall. A Raspberry Pi touchscreen with our shared Google Calendar, live scores for the teams we follow, the ceiling lights, a grocery list you can add to from your phone, and a few games for when someone is bored.

    Python / JavaScript / Raspberry Pi / Zigbee

  6. 4

    Object-tracking launcher 2024

    A self-aiming ping-pong ball launcher. It runs YOLO object detection and combines depth, color and motion data to follow a target, moves its servos in real time and fires using pressurized oxygen.

    Python / YOLO / Computer vision / Circuits

  7. 1

    Measuring marine mammals from photos 2023

    Co-author on a doctoral thesis about a program that works out the size of marine mammals from a single image. Published in Marine Mammal Science.

    Python / Computer vision / Research

Substitutes

  1. 12

    Tic-Tac-Toe AI 2023

    A Monte Carlo Tree Search agent, built for a master's course at Carnegie Mellon's Heinz College.

    Python / Algorithms

Transfer history

Where I've worked.

SeasonClubRole
Jun – Aug 2026AstraZenecaBuilt a department-wide internal tool in Python with Dash that benchmarks how well different AI tools find relevant research publications for a given set of criteria.AI Strategy & Innovation Intern
Jun – Aug 2025Denver VenturesResearched and presented market and competitive analysis across the technology sector, and supported due diligence on early-stage startups.IR & Marketing Intern
Aug 2024 – May 2025Stratus Technology Integration ServicesHigh school capstone as a subcontractor to General Dynamics IT. Collected software vulnerability data with SQL and visualized it to find trends.Database Intern
Jun – Aug 2022Arjuna SolutionsBuilt models that helped nonprofits decide how much to ask donors for and when, using donor datasets.ML Developer
Jan 2022 – Sep 2023Sugar LabsAdded synthesizer support to Music Blocks, a music-and-math platform for kids, so it can play single notes and scales.Open-source contributor

Honors