Diya Singh — art ∩ logic

§ 01

Abstract

Most people pick a side: the studio or the lab. I kept both keys.

I'm an M.S. Computer Science student at Purdue University (B.S. CS '26, M.S. '27) focused on artificial intelligence, machine learning, and deep learning. My undergrad concentrations run from machine intelligence to computer graphics to systems software — and my minors in art & design and mathematics are not a contradiction; they're the method.

The pattern in my work: take something messy and human — 2 million walkability reviews, a company's email triage, a card game's luck — and give it structure a machine can reason about, then give the machine an interface a human actually wants to touch. I've shipped that loop at Envita Solutions, FractalHive, Haleon, and in research at Purdue, and I teach it every Friday to a lecture hall of 400 as Head TA of CS180.

§ 02

Selected works

each figure is plotted live by the algorithm named beneath it — click a drawing to re-roll its seed

fig. 1 — rose curve, r = cos(kθ)

may 2026featured

Bloom — AI-powered PCOS management ↗

A cycle-aware iOS app (React Native + Expo) that uses Gemini 2.5 Flash to generate daily meal and workout plans tuned to PCOS phase, pantry stock, and real equipment. Cycle prediction via exponential-decay-weighted averaging over six prior cycles with confidence intervals; imports Flo data; one daily AI call with a graceful static fallback so it works offline and inside free-tier limits.

React Native
Expo
Gemini 2.5
iOS
health

fig. 2 — detections under corruption

dec 2025

Animal detection under corruption ↗

Fine-tuned and benchmarked YOLO, Mask R-CNN, and RT-DETR on 9 animal classes from Open Images under Gaussian blur and JPEG compression. RT-DETR won: best baseline mAP@0.5 (0.788) and the smallest high-blur degradation (−11.2% vs −17.7% for YOLO).

PyTorch
RT-DETR
YOLO
Mask R-CNN

fig. 3 — expected-value tree

dec 2025

Flip7 AI agent ↗

A dynamic-programming hit/stay agent for the Flip 7 card game: memoized, deck-weighted expected values at every state, action by argmax. Pivoted from Q-learning after diagnosing multi-agent reward instability — the DP agent matched human win rate (50%) over 182 evaluated rounds.

Python
dynamic programming
game theory

fig. 4 — gaze trace, lissajous

sep 2024HackMIT finalist

ReadCoin ↗

A reading app that watches you read: real-time eye-tracking drives auto-scroll and pays out incentives per line read. React + Python over WebSockets, built in 24 hours, finalist at HackMIT.

React
Python
WebSockets
eye tracking

fig. 5 — throughput traces

apr – may 2025

Live streaming system

Real-time streaming in Docker containers on a public GCP VM. Automated pcap capture and analytics across network conditions in Python, with statistical reports and matplotlib/seaborn visualizations.

Docker
GCP
Python
networking

fig. 6 — L-system, 7 generations

2025

Procedural 3D tree ↗

Trees grown from grammar: L-systems and turtle graphics in C++17/OpenGL with a real-time turntable, ImGui controls, and procedural leaves and flowers — no textures, only rules.

C++17
OpenGL
L-systems
graphics

§ 03

Chronology

internships & research, newest first

may – aug 2026now
AI & ML Automation Intern — Envita Solutions
- Replatformed an AI email classification & extraction pipeline that auto-creates customer request records and routes them to the database.
- Pushed accuracy from 92% to >95%, beating the prior Power Automate flow and manual scheduling.
- Replaced the flow with an auditable system that cuts manual triage and surfaces edge cases for human review.
- LLMs
- classification
- automation
jul – sep 2025
Product Development Intern — FractalHive
- Delivered a responsive audit web app for an education-sector client, cutting manual processing time 30%.
- Designed an end-to-end schema with validation and multi-level access control; built role-specific dashboards for relationship managers and underwriters in a B2B commercial-banking platform.
- React
- SQL
- B2B SaaS
jan 2025 – present
Interdisciplinary Research Intern — DUIRI, Purdue
- Engineered an AI walkability assistant informed by LDA topic modeling over 2M+ user reviews in Python.
- Led user-centered design and prototype testing; feedback integration lifted engagement 25%.
- Manuscript in preparation: "AI-Driven Walkability Ecosystem: Enhancing Public Health through Personalized, Social, and Adaptive Mobility Solutions."
- NLP
- LDA
- UX research
jul – aug 2024
Machine Learning Intern — Haleon · Centrum
- Architected an end-to-end facial-recognition diagnostic for nutrient deficiency — system architecture, Firebase schema, and the ML pipeline, in Flutter and Python.
- Agile sprints with weekly stakeholder feedback improved feature accuracy 20%.
- computer vision
- Flutter
- Firebase
may – jul 2024
Machine Learning Intern — NRoad
- Built a finance-document parsing model and REST API on a microservice architecture with FastAPI, SQLAlchemy, and MySQL.
- FastAPI
- SQLAlchemy
- MySQL
aug 2023 – jan 2024
Undergraduate Researcher — Dr. Wenzhuo Wu's Lab, Purdue
- Built TinyML models for wearable sensors tracking athletic performance and recovery.
- Modeled a digital screen printer in Unity, cutting resource waste 30% and testing time 35%; presented at the Purdue research expo.
- TinyML
- Unity
- wearables
jun – aug 2023
Machine Learning Intern — Indium Software
- Benchmarked 10 Hugging Face LLMs for finance and healthcare; the evaluation report improved model-selection accuracy 25%.
- Hugging Face
- LLM evaluation

§ 04

Toolbox

languages

Python
Java
C / C++
SQL
JavaScript
MATLAB
Bash

ml & data

PyTorch
TensorFlow
scikit-learn
pandas
numpy
nltk
matplotlib
seaborn
Hugging Face

apps & apis

React
React Native
Flutter
FastAPI
SQLAlchemy
Firebase
Supabase
JUnit

cloud & tools

Docker
GCP
Vertex AI
Azure ML
Git
Unity
Figma
Claude Code

certifications ↗

ML Specialization — DeepLearning.ai / Stanford
Deep Learning — NVIDIA
Intro to ML — Kaggle
AWS AI & ML Scholar

§ 05

Education

m.s. · 2025 – 2027

M.S. Computer Science

Purdue University, West Lafayette

Artificial intelligence · machine learning · deep learning

algorithmic game theory — data mining — artificial intelligence — data engineering I — parallel computing — theoretical CS

b.s. · 2022 – 2026

B.S. Computer Science

Purdue University · minors in Art & Design and Mathematics

Machine intelligence · computer graphics · systems software · algorithmic foundations

data structures & algorithms — advanced algorithms — operating systems — compilers — systems programming — computer networks — computer graphics — probability — linear algebra — numerical methods

§ 06

Teaching

purdue cs · aug 2023 – present

CS 180

Head Undergraduate TA — Problem Solving & OOP now

Six semesters with CS180 — three as undergraduate TA, one as Student Success TA, and two as Head TA. Teaching Friday classes to 400+ students; leading labs, grading, and coordinating student concerns. Mentoring on debugging, optimization, and engineering practice in and out of office hours; as Student Success TA, provided individualized academic guidance and exam preparation.

CS 381

TA — Analysis of Algorithms spring 2026

Office hours and help sessions on asymptotic analysis, dynamic programming, graph algorithms, and proof-based reasoning; proctoring and grading for the upper-level course.

CS 182

TA — Discrete Mathematics spring 2024

Problem-solving technique and theory support, office hours, and grading.

§ 07

Honors & involvement

Ford Motor Company Scholarship — CS Corporate Partners, Purdue · 2024 & 2025
Pillars of Excellency Scholarship — Teaching & Scholarship, Purdue · 2024
BoilerMake Hackathon — Design Team Lead, Executive Board · feb 2023 – present
Women in Science Program — Jandos Scholar, mentor, social media committee · aug 2022 – present
Publication in preparation — AI-Driven Walkability Ecosystem, DUIRI, Purdue
Member — Rewriting The Code · CodePath · Amazon Campus Prep Series '25

§ 08 — correspondence

Let's make something
that thinks and feels.

Open to ML/SWE opportunities, research collaborations, and good conversations about either circle of the Venn diagram.

singhdiya2004@gmail.com GitHub LinkedIn Résumé