DistPI: Hybrid RAG System & Semantic Routing Protocol
I designed a distributed RAG system that efficiently routes queries across document shards using semantic-based routing. The system intelligently synthesizes answers by combining distributed document search with LLM reasoning, achieving optimal query routing while minimizing network overhead.
ClosedAI: GPT-2 Inference Engine & CUDA Optimization Suite
I built a high-performance GPU-optimized inference engine for GPT-2 that prioritizes throughput and memory efficiency. Through custom CUDA kernel development and model quantization, the engine achieves significant speedups and memory reductions while maintaining model quality.
My Hub of Academia: University Application CRM
I built a full-stack university application tracking system designed to handle concurrent users with a robust, scalable architecture. The system prioritizes data consistency and query performance, combining a React frontend with a carefully optimized relational database backend.
GrindMate: Collaborative Study Platform
I developed a collaborative study platform where students can schedule sessions, chat in real-time, and generate study materials with AI. The system balances performance and consistency through a dual-database architecture and emphasizes developer experience with strong testing and type safety.
Open Seizure Detection
I led a 6-engineer team to build a real-time seizure detection system for Android that processes wearable sensor data with minimal false alerts. The system prioritizes both reliability and accessibility, ensuring vulnerable users can depend on accurate, timely notifications.
Spring 2024 Data Science Projects
This repository contains various data science projects I have completed in Python over the course of the Spring 2024 (my Junior Spring) semester at Johns Hopkins
University. My projects derive from two of my courses: Computational Stem Cell Biology (CSCB) and Information Retrieval and Web Agents (IRWA).
Computer System Fundamentals
This course covers the design and performance of computer systems, ranging from simple 8-bit microcontrollers to 32/64-bit RISC and x86 CISC architectures.
It starts with logic gates and digital circuits, progressing to topics like arithmetic and logic units, registers, caches, memory, and pipelined execution.
The course also includes instruction set architectures, interrupts, and peripheral communication protocols, with practical programming projects in assembly
language and processor simulators.
Fall 2023 Data Science Projects
This repository contains various basic data science projects I have completed over the course of the Fall 2023 (my Junior Fall) semester at Johns Hopkins University.
The majority of my projects were completed as part of the Biomedical Data Science course along side its accompanying Biomedical Data Science Laboratory labeled under the BME-DS folder.
These projects range from basic classification tasks using only the Sci-Kit Learn library to more advanced image classification tasks using convolutional neural networks using Pytorch.
On the other hand, the AI folder contains my final project for the Artificial Intellgience course. This project measures the effectiveness of various neural network model configurations
on a standard classification task using Pytorch.
K-RCPS
K-RCPS is a high-dimensional extension of the Risk Controlling Prediction Sets (RCPS) procedure that provably minimizes the mean interval length by means of a convex relaxation.
Under the supervision of JHU CS Professor Jeremias Sulam and graduate student Jacoppo Teneggi, I worked towards implementing the K-RCPS algorithm in HuggingFace and publishing it
for official use as part of the HuggingFace Transformers library.
Coursera-JHU Data Science Specialization
This repository contains various projects I have completed as part of the Coursera JHU Data Science Specialization. The specialization consists of 10 courses that cover the
fundamentals of data science, including data manipulation, data visualization, machine learning, and more.
