Built an English-to-SQL system using schema-aware prompting, LoRA fine-tuning, and execution-guided decoding to reduce query errors by 25%. Deployed the model as a FastAPI REST API for seamless application integration.
An AI-powered web app that generates personalized, age-appropriate bedtime stories based on a child’s name, age, and interests. Built with Next.js, React, and Tailwind CSS, it combines real-time content generation with safe, engaging narratives for an enjoyable storytelling experience.
DataSage is a full‑stack, web‑based analytics platform that automates data profiling and summarization. It intelligently detects data types and bins variables to instantly render survival curves, histograms, and categorical plots. This app helps streamline early‑stage exploratory analysis with interactive, real‑time visualizations.
Built using Groq, LangChain, and Streamlit, this tool helps businesses connect with potential clients by extracting job listings from a company's careers page and generating customized cold emails.
Financial markets are dynamic and complex, requiring quick, data-driven decisions. This project involves developing an AI agent that can autonomously make trading decisions based on real-time data analysis, market sentiment, historical patterns, and predictive models. Agentic AI refers to AI systems that act autonomously towards goals, making this ideal for trading where timing and strategy are critical.
Developed an AI-powered system that recommends books based on natural language queries. The system leverages large language models (LLMs) and vector search to provide book recommendations based on themes, emotions, and classifications.
Developed a YouTube script generator using GPT-4 to automate the creation of engaging video content by generating titles and scripts based on user-defined prompts.
Implemented a YOLOv5-based Traffic Sign Detection Model using deep learning to accurately perform real-time detection of four distinct traffic sign classes: Traffic Light, Stop, Speed Limit, and Crosswalk.
Built a model that can predict the NBA's Most Valuable Player using Linear Regression, Ridge, Random Forest, and Boosted Trees.
Performed Time Series Analysis on monthly weather patterns in Downtown Los Angeles using the SARIMA and Lagged Regression methods. The purpose of this study is to gain insights into the presence of global warming, understand seasonal variations, and identify any significant trends or patterns in the weather data.
Created an interactive sales dashboard in Tableau to explore the yearly history of video games through the use of various parameters and filters. The dashboard is able to sort by geographical region, time range(years), publishers, video game titles, and much more.
I delve into the World Happiness Report 2023, a rich dataset offering insights into the factors influencing subjective well-being across diverse countries. This approach aimed to provide a better understanding of the factors impacting individuals' life satisfaction in North America, and to uncover meaningful patterns.
Performed Data Analysis on the Global YouTube Statistics 2023 Dataset to understand factors influencing the success of top YouTube channels worldwide. I explore the relationships between subscriber counts, video views, highest yearly earnings, and explore the possibilities of creating a machine learning model that can predict any channel's subscriber counts.