I'm a Computer Science and Machine Learning student at the University of Maryland, passionate about architecting and building intelligent, scalable systems. I'm driven by the challenge of solving complex, real-world problems with code, which I explore through my projects in AI, distributed systems, and performance optimization. As the Vice President of BigThink AI, I also enjoy leading workshops and sharing my knowledge with the tech community. I'm also engaged in several concurrent research projects in Computer Vision, MultiModal ASR + Sentiment Analysis, and Circuit Stability + Mechanistic Interpretability. Stay posted for updates!
Note: These repositories represent academic research and experimental prototypes exploring the limits of distributed systems and AI. They are not active commercial products.
- Building Resilient AI Systems: Exploring fault-tolerant, high-throughput architectures using microservices, Go, Kubernetes, and distributed consensus algorithms.
- Developing Advanced AI Agents: Creating multi-agent systems and autonomous research agents that leverage Retrieval-Augmented Generation (RAG) for dynamic, fact-based reasoning.
- Engineering High-Throughput ML: Researching graph-aware anomaly detection approaches and exploring optimizations in time-series analysis and machine learning inference.
I explore applying AI to solve problems across various domains. My work spans the entire stack—from system-level optimization to immersive 3D UX—across these distinct research areas:
| 🏢 Vertical / Domain | 🚀 Research Focus | 🛠️ Tech Stack |
|---|---|---|
| 🏥 Healthcare | Medical diagnostics & microservice architectures | Go, K8s, LangGraph |
| 💳 FinTech & Markets | Predictive modeling & sentiment analysis | XGBoost, FinBert, Next.js |
| 🚚 Supply Chain | Event-driven logistics & inventory tracking | Java Spring, SQS, DynamoDB |
| 🛡️ Security & Infosec | GAN-based detection & DevSecOps tooling | FastAPI, TensorFlow, GKE |
| 🎙️ Audio & Speech | Speaker diarization & sentiment extraction | WhisperX, Python, Signal Proc. |
| 🧪 Multi-Agent Systems | Swarm intelligence & reasoning orchestration | Consensus Algos, GRPO |
| 🎮 Immersive & Gaming | Real-time metaverse & 3D visualization | Three.js, WebSockets, React Native |
| 🎓 EdTech | AI-driven pedagogy & remediation | React, LLM APIs, Java |
| 🛠️ DevTools & Ops | Agent observability & code review automation | C#, .NET, ClickHouse, OpenAI, Python |
| 🛍️ E-Commerce | Conversion optimization & UX research | Next.js, Gemini 2.5, React |
| 🔬 Deep Research | Autonomous reasoning & Tree-of-Thoughts | LangChain, Milvus, Python |
| 🧠 Wellness & HCI | Life orchestration & sentiment visualization | Pgvector, Three.js, Transformers |
| 🗣️ NLP & RAG | Text-to-SQL agents & document synthesis | Python, Jupyter, Mistral |
| 👁️ Computer Vision | Video segmentation & edge-based inference | PyTorch, CUDA, SAM2 |
| 🚀 Rapid Prototyping | Experimental frameworks and MVPs | Python, Selenium, Flask |
| Project | Description | Key Technologies |
|---|---|---|
| Research: Anomaly Detection Systems | Exploring context-aware detection approaches for distributed systems. Investigating time-series metrics and graph topology methods for anomaly detection in production environments. | Python, Polars, XGBoost, Graph Theory, Scikit-Learn |
| MedInsight: AI-Powered Medical Diagnostics | A microservices platform with a Go API gateway and Python AI service, orchestrated on Google Kubernetes Engine (GKE). Designed a decoupled system using Google Pub/Sub for fault tolerance. | Go, Python, GKE, Pub/Sub, Docker, Gemma 2 |
| AuraFlow: Enterprise Supply Chain Intelligence | A cloud-native, event-driven microservices system in Java (Spring Boot) on AWS. Implemented the Saga pattern with AWS SQS for data consistency across microservices. | Java, Spring Boot, AWS, Microservices, SQS, DynamoDB |
| AzureDev Nexus: Intelligent Developer Ecosystem | A platform in the .NET ecosystem, featuring an AI-powered code reviewer using Azure OpenAI and a real-time collaboration platform with SignalR. | C#, .NET, Blazor, Azure OpenAI, SignalR |
| Connect Verse: Real-Time Social Metaverse | A mobile-first, real-time social app with a Node.js backend managing concurrent WebSocket connections. Implemented a Redis caching layer to reduce database load. | React Native, Node.js, WebSocket, Redis, PostgreSQL |
| Multi-Agent Pet Store on AWS | A multi-agent system using AWS Bedrock Agents. Designed a text-to-SQL pipeline with a self-correcting agent loop for natural language database interaction. | AWS Bedrock, Lambda, OpenSearch, RDS |
| LLM Fine-Tuning with GRPO | Fine-tuned Microsoft's Phi 14B model for mathematical reasoning on the GSM8K dataset. Implemented the GRPO algorithm with custom reward functions to improve complex reasoning. | PyTorch, TRL, Unsloth, Hugging Face |
| Project | Description | Key Technologies |
|---|---|---|
| DreamWeft | An experimental tool for visualizing abstract concepts through generative art and ambient computing. | Next.js, TypeScript, GLSL Shaders, Framer Motion, Tailwind |
| Tempo | A time and energy management dashboard designed to provide a holistic view of personal productivity. | Next.js, TypeScript, Recharts, Framer Motion, Tailwind, shadcn/ui |