CV

Bachelor of Science

University of Arizona

Computer Science

• Algorithms
• Databases
• Automata
• Compilers
• Linear Models
• Machine Learning
• Computer Vision

Software Development Engineer

Amazon

Lead engineer optimizing transport network speed and efficiency at scale. Delivered 4 minor and 2 major projects in 5 months, resulting in SDE-2 promotion recommendation. Led cross-team workflow design impacting 230MM+ configs and designed a global AWS/Spring email service processing 1000+ emails/min.

• SDE-2 promotion recommendation after 5 months
• Tech lead for transport network optimization (impacting 230MM+ configs)
• Designed AWS/Spring email service (1000+ emails/min, 5+ teams, global vendors)

Research Engineer

University of Arizona

Designed deep RL transformer math OCR models, achieving a 6% increase over SOTA using PyTorch and Stable Baseline. Overhauled MathML parser with Nom in Rust, improving traceability for 30% of tag syntax errors during AST generation.

• 6% improvement over SOTA in math OCR
• Improved MathML parser traceability with Rust (30% more errors traced)

Software Development Engineer Intern

Amazon

Built and deployed a React/TypeScript full-stack internal tool to CloudFront with proxy-based authentication, enabling secure monitoring of millions of purchase orders globally. Automated data transformation pipeline with AWS CDK, DynamoDB, Kinesis, Lambda, S3, Glue, Athena, IAM, achieving 98% improvement in data availability.

• Enhanced operational transparency across 2 orgs
• 98% improvement in data availability (AWS pipeline automation)

Lead Software Developer

Tech Core

Led web and AR/VR projects for 10+ clients across 6+ university departments and startups. Optimized SQL queries and JS through code profiling, improving web app load times by 80.51% across 3 services.

• Led projects for 10+ clients (6+ departments/startups)
• 80.51% improvement in web app load times (SQL/JS optimization)

Undergraduate Research Assistant

University of Arizona

Developed modular procedural generation framework in C++ for RL environments. Second author on AAAI paper. Sole recipient of Research Excellence Award for CS and 2x Galileo Circle Award.

• Second author on AAAI FSS 2021 paper
• Sole recipient: Research Excellence Award (CS), 2x Galileo Circle Award

Galileo Circle Scholar

University of Arizona

$1000 scholarships awarded to students demonstrating exceptional potential in sciences.

Academic Merit Scholarship

University of Arizona

$120,000 competitive undergraduate tuition scholarship.

Summa Cum Laude

University of Arizona

Awarded to students graduating with a GPA of 4.0 over 4 years.

Highest Academic Distinction

University of Arizona

4-time recipient for maintaining a 4.0 GPA for an academic year.

Excellence in Undergraduate Research

University of Arizona

Awarded for significant contribution, originality, creativity, and independence in research.

Dean’s List With Distinction

University of Arizona

8-time recipient for maintaining a 4.0 GPA for an academic year.

Host Management System

Architected an EC2 host management system leveraging cryptographic techniques for secure user authentication and session management. Features include host monitoring, dynamic fleet scaling, and controlled termination. Used encryption protocols to safeguard against man-in-the-middle attacks.

• Python, Apache, AWS Boto
• Encryption protocols for security
• Dynamic scaling and monitoring

TaylorGPT

Designed a 1.4M parameter character-level transformer decoder model to generate Taylor Swift lyrics from scratch using PyTorch. Built custom masking, multi-head attention, transformer blocks, and enabled multi-GPU training.

• PyTorch, Lightning, Numpy
• Custom transformer implementation
• Multi-GPU training

Tau Compiler

Developed an LL1 Recursive Descent compiler for Tau, including lexer, parser, decorator, and generator with syntax analysis, scoping, type checking, optimization, and memory management.

• Python, PyTest
• Full compiler pipeline

Cat(G)AN

Implemented a GAN with 3.5M parameters from scratch in NumPy, TensorFlow, and Keras to generate cat images. Developed custom training and evaluation pipelines.

• Python, TensorFlow, Keras, Matplotlib
• Custom GAN architecture

Start Journey

Designed a full-stack stable diffusion-based image generator with PyTorch, CUDA, HuggingFace, and FastAPI for the backend and React in TypeScript for the frontend. Enabled scalable image generation and user-friendly web interface.

• Python, PyTorch, HuggingFace, CUDA, FastAPI, React
• Full-stack ML deployment

ResumeGPT

Designed and deployed a Streamlit application with LangChain, ChatOpenAI, and ChromaDB that uses LLMs to answer questions about me, including customized vector DB logic and interactive UI.

• Python, PyTorch, Streamlit, ChromaDB
• LLM-powered Q&A

Signature

Designed a multi-platform desktop application for e-signature extraction from photographs using custom image recognition logic. Combined normalization and thresholding for robust signature extraction.

• Python, OpenCV, Pillow, Matplotlib
• Image normalization and thresholding

AI-PlaysSnake

Implemented Snake and a genetic neuro-evolving RL model to beat it, including a custom physics engine and game graphics.

• Python, NEAT, PyGame
• Genetic RL for games

Vector Quantized Reinforcement Learning

Research in progress

Applying vector quantization to reinforcement learning. Developing a framework for learning prototype vectors in feature space. Initial experiments show promise with a cosine similarity loss-based approach.

Transformer-based OCR for Mathematical Expressions

In development

Developing a transformer-based OCR model to read mathematical expressions from images and convert them into grammatically correct LaTeX. Hybrid two-stage learning: transfer learning followed by reinforcement learning. Uses an ad-hoc AST for loss in a context-sensitive, Turing-complete language.

Modular Procedural Generation for Voxel Maps

AAAI Fall Symposium

Pyarelal, A., Banerjee, A., Barnard, K. (2022). Modular Procedural Generation for Voxel Maps. In: Gurney, N., Sukthankar, G. (eds) Computational Theory of Mind for Human Machine Teams. AAAIFSS 2021. Lecture Notes in Computer Science, vol 13775.

Winner

MLH: To the Moon and Hack

Received In-Flight Anomaly prize among 330+ contestants for Project Space Station, an AI trained via neuroevolving genetic RL.

Winner

MLH: Hack Arizona

Won Amazon’s Most Impactful Use of Data-Driven Technology for developing a machine learning model for distracted driver detection with AWS Sagemaker, among 200+ contestants and 50+ teams.

Mentor

MLH: Hack Georgia Tech

Mentored hacker teams at Hack Georgia Tech Hackathon with 2500+ attendees, supporting innovative project development.

Founder & President

Google Developers Student Club

Founded and led the GDSC at University of Arizona, organizing 30+ workshops for 180+ members and collaborating with 50+ international chapters. Drove innovation and engagement, increasing CS club participation by 132%.

Winner

Google Developer Challenge

2nd place for solving algorithm problems at Google Tech Challenge among 50+ contestants.

Volunteer

Blue Chip Leadership

Participated in community service projects with local AIDS and sustainability organizations, including SAAF and Tucson Trees Please.

Council Member

University of Arizona Residence Hall Association

Promoted sustainability practices and awareness across campus. Organized and voted on dorm and RHA programs, events, budgets, and issues. Nominated twice for RHA Person of the Week.