π M.S. Information Systems @ Northeastern University
π
AWS Certified Solutions Architect | Azure Fundamentals Certified
πΌ 5+ Years Experience in Full-Stack Development & Cloud Engineering
π¬ Published Researcher in Secure Data Communication & Spatial Intelligence
π Building Scalable AI Solutions & Cloud-Native Applications
π» Generative AI developer at Humanitarian AI/Stellis Labs where I work on developing LLM-based by fine-tuning LLMs.
π Lead Research Assistant | Northeastern University. Handpicked by Prof. Kelly Wellman and Prof. Amy Pei to spearhead two high-impact ML/AI projects, delivering cutting-edge AI solutions to professors and key stakeholders. Played a pivotal role in advancing AI-driven research and large-scale data analysis.
This is a specialized e-commerce platform tailored for artisans and craft sellers. Unlike mainstream marketplaces that are cluttered with various products, HCS is exclusively designed to showcase and promote authentic handcrafted items, providing a dedicated space for genuine creators to market and sell their work. ππ’ π£ Let's collaborate if you wish to work on this with me! ππ’ π£
Domain-Specific Vector Embedding Optimization: Fine-tuned embedding models on 5000 (synthetic cryptocurrency data) domain-specific data using contrastive learning, RAG-based dataset synthesis, and synthetic data generation, improving retrieval accuracy and performance.
Tech Stack: Hugging Face Transformers, RAG, Llama 3 - 70B (Groq), ChromaDB, all-MiniLM-L12-v2, mxbai-embed-large, Python.
Developing a centeralized solution for "on a given problem statement, recieving an array of solution strings as input, then intelligently merging them into a single, coherent solution while adhering to LLM token limits." **Solution: ** Utilizing Hierarchial Processing Pipeline strategy to Chunk, Summarize using HuggingFace embedding models. GitHub Issue GitHub Repo
LLM-powered meeting summarization tool using custom prompt engineering to extract key takeaways, action items, and insights. Built with GPT-4, LangChain, and Streamlit. Enables faster, clearer understanding of meeting content. Audio/Videos processed to generate transcript (pipeline) built using WhisperAPIs.
AI-powered patient prioritization using RAG architecture with Groq and ChromaDB. Assigns triage levels based on severity with real-time processing. YouTube
This project showcases how Retrieval-Augmented Generation (RAG) can be implemented using LangChain and FAISS for building a context-aware chatbot. Integrates document loading, vector store indexing, and query processing for efficient response generation. Streamlined for easy deployment and extensibility using GCP Vertex AI. YouTube
This project demonstrates how EKS (kubectl) with Helm, Terraform provisioning can be used to deploy a Machine Learning model. Configures VPCs and developed APIs to fetch downstream data. Medium Article
A library that showcases the books they have! Demonstrating backend service development using Java, Springboot, Gradle, Docker, Redis, Jenkins (CI/CD).
Kubernetes-powered pharmacy platform handling 5k+ daily transactions with Stripe payments, real-time inventory tracking, and automated delivery scheduling.
Real-time sensitive data protection using TFAutoModelForTokenClassification, deployed on GKE with Grafana monitoring.
π¦οΈ AirCast - Air Quality Prediction
LSTM-based platform predicting air quality with 0.8 RMSE, featuring real-time data pipelines and a React dashboard.
Scalable weather analytics platform processing 1TB+ of SEVIR/GOES/NEXRAD data with Airflow ETL pipelines and Streamlit dashboards.
AI-driven dining platform with Gemma:2B chatbot for ordering, Snowflake analytics, and personalized menu recommendations.
CUDA-accelerated cancer prediction system achieving 98.7% accuracy with hybrid Dask/DDP pipelines and multi-GPU parallelism.
Demonstration of a real time event driven communication system for 2 applications: main app and admin app. Admin app can post products and main app which is for users can like the product. Tech Stack: RabbitMQ, Angular.js
High and Low Level System Design:
π Check my Algorithm Repository