I like taking projects from a rough idea to something people can actually use, whether that is
a model, a robot, or a web app. Most of what is here started as a question I was curious about.

_{PYTHON  ·  C++  ·  JAVASCRIPT  ·  TYPESCRIPT  ·  NEXT.JS  ·  REACT  ·  PYTORCH  ·  SUPABASE  ·  ARDUINO  ·  LATEX}

  Mental Health Sentiment Analysis: hybrid BERT and Random Forest pipeline for automated mental health text triage (about 85% accuracy), deployed on Hugging Face Spaces with a GitHub Pages frontend. The basis for a paper submitted to IJHSR.
  Voice Health: App that tracks the long-term progression of voice-impacting degenerative diseases (aphasia, Parkinson's, etc.).
  Parkinsons TinyWatch: an Arduino-based hardware companion for Parkinson's-related signal sensing.
  Pokémon Showdown AI: a supervised imitation-learning AI for competitive Pokémon Showdown (Gen 9 OU), built as a residual MLP trained on a 725-dimensional feature vector from my own gameplay data, collected through a terminal proxy bot using poke-env.

I've also built an XGBoost and SHAP pipeline on the UCI Parkinson's voice dataset as a technical assessment for a university research lab application, and wrote an AP Research paper on implicitness and adaptability in gamified educational environments, with statistical analysis of the game metrics behind it.

I compete with VEX Robotics Team 285C.

  VEX V5 Template: template for VEX V5 programming build on PROS & LemLib.
  Pushback 285C: competition code for the Pushback season for 285C Discobots Chaos.
  Override 285C: competition code for the Override season for 285C Discobots Chaos.

Outside competition code, I designed automatic motion-chaining formulas for LemLib's moveToPoint() and turnToHeading(), using kinematic stopping distance and trapezoidal or triangular motion profiles so the robot carries its current speed into the next movement instead of stopping and restarting between points.

  Project 1600: a full-stack SAT prep and mentoring platform built with Next.js and Supabase.
  IAGE 2D Platformer: an original 2D platformer game used for a research project over implcitness & adaptability in gamiefied education.
  APUSH Practice: a gamified AP US History practice site.

  Mental Health Sentiment Analysis: wrote a paper on a hybrid BERT and Random Forest pipeline for mental health text triage, submitted to the International Journal of High School Research (IJHSR).
  Rice University DHYS: participated in the Digital Health Young Scholars program at Rice University, where I created a watch to address Parkinson's disease in underserved areas.
  Research lab technical assessment: built an XGBoost and SHAP pipeline on the UCI Parkinson's voice dataset as part of a university research lab application.
  AP Research: independent paper on implicitness and adaptability in gamified educational environments, including original statistical analysis of in-game metrics.
  VEX Robotics: lead programmer for team 285C, competing in V5 competition.

_{This updates as projects wrap up or new ones start. The repos are the source of truth.}

More at ayaangoswami.dev