Explore my hands-on work with modern tech stacks, from AI/ML to full-stack web apps.

A personalized AI assistant powered by a large language model capable of answering questions about my background, experience, projects, skills etc.

An AI-assisted discovery platform to recommend content based on user mood using a hybrid of semantic search and programmatic filtering.

A backend ingestion pipeline for my personal assistant, to process all documents that describe my work, projects, background, etc. that power semantic search on my portfolio.

A dedicated backend service that orchestrates prompt routing, retrieval, tool use, and workflow logic for Jarvis.

A backend data ingestion pipeline for MoodFlix, an AI-assisted discovery platform, for search and recommendations.

A resilient, multi-threaded data pipeline designed to ingest high-frequency data from GitLab’s GraphQL API into PostgreSQL for downstream analysis.

A simple card swipe memory game built end-to-end completely using prompts with bolt.new

A simple card pairing game built end-to-end completely using prompts with bolt.new

A simple bookstore application built using Java EE, Apache Derby, Apache Tomcat, Apache Axis, and Jersey.

Modeled a supervised learner using linear regression with risk minimization techniques to illustrate the concept of overfitting (a doomed learner) using graphs.

Built a classification learning model to classify web pages into categories by extracting words from a source of hand classified web page.

An app on the google assistant to help York University students find information like GPA, courses enrolled, etc. quickly using voice-recognition.

A real time sentiment analysis big data pipeline of top trending topics on Twitter using Apache Spark and HDFS.