OVERVIEW
$> whoami
I'm pursuing my Master's in Computer Science at Northwestern University, focused on GPU systems and machine learning. I've received my B.S. in Computer Science from Rutgers University Honors College.
I have experience working on production software systems in industry settings, including machine learning pipelines, LLM evaluation frameworks, backend APIs, and data infrastructure.
My passions sit at the intersection of systems and machine learning, which I hope to explore further at Northwestern University. I'm especially interested in working on optimizing GPU performance of ML models by bypassing high-level abstractions to engage directly with performance and hardware efficiency.
$> interests
LLM SystemsParallel ComputingGPU OptimizationRL / RoboticsReinforcement Learning
Northwestern University
Rutgers University – Honors College
EXPERIENCE
TeachShare
- ▸Orchestrated a unified multimodal grading pipeline with rubric-aware prompting and structured feedback generation
- ▸Engineered HTML parsing and chunking to reduce LLM token usage on large Polotno JSON inputs
- ▸Developed backend services and evaluation workflows using Solid.js, Vertex AI, and Zod schema validation
- ▸Implemented chained multi-step tool-calling for an LLM-powered chat system supporting dynamic task execution
J.P. Morgan Chase
- ▸Developed and deployed an LLM Evaluation & Regression Testing framework for internal teams with support for Phoenix and Langsmith
- ▸Provided an asynchronous modular pipeline with trace-level observability and LLM-as-judge evaluation for firm-wide LLM workflow benchmarking
- ▸Built a generalized document Q&A pipeline for high-volume workflows (RFPs, compliance) via procedural chunking and retrieval-augmented generation
Hone Health
- ▸Developed a custom content-injection Azure DevOps extension for processing ticket similarity with vector embedding search to display similar work items
- ▸Engineered Azure ETL pipelines to ingest Customer.Io webhook data to Azure Blob Storage into KQL eventhouse
- ▸Architected a migration pipeline for bulk SQL scripts to transfer staging data to a production history database
Depton LLC
- ▸Engineered a distributed web scraping framework for real-time stock ticker aggregation, processing 2k data points daily.
- ▸Led the end-to-end development of the company’s main internal web platform using Next.js, including authentication, API integration, component design, and deployment
Rutgers University
- ▸Assisted with the development of a pose estimation robotic arm control system
- ▸Managed puppeteer scripts for automatic inventory document management
VEX 750B Robotics Team
- ▸Designed a modular robot control framework with PID motion tuning, Odometry, and Pure Pursuit path tracking.
- ▸Programmed a completely autonomous skills routine which got 1st place in the NJ State competition of 50+ teams, qualifying for the VEX Worlds Competition for the first time in 4 years
- ▸Developed a React Native scouting app with optimized heuristics to predict match performance from live telemetry data
PROJECTS
ALLOCATED
CUDA GPT
GPT-style Transformer built in raw C++/CUDA
GPU-Accelerated SQL Query Engine
GPU-accelerated SQL query engine for large-scale data processing
DebugERR
Fine-tuned LLM for automatically detecting errors and providing actionable fixes
MeLM
Fine-tuned LLM replicating personal style
VEX 750B
Competition robot with autonomous path tracking
SNN Quadruped Locomotion
Diffusion-generated quadruped locomotion with SNN temporal extractors
Reflevapp.com
AI journaling with mood intelligence
Kanzure.com
Modern productivity suite for integrated task management
Kernel Configuration Search
Automated CUDA kernel performance exploration
Mentalth
AI-powered Discord moderation bot for mental health
Neat.js
NeuroEvolution of Augmenting Topologies from Scratch in JavaScript
TECHNICAL
SPECS
ESTABLISH
CONNECTION
Open to internship and full-time opportunities in AI engineering, systems, and full-stack. Always up to talk compute, CUDA, or LLMs.