Data Science / Development Portfolio

P01

Psychiatric Query RAG

In development: An RAG system that provides answers to psychiatric questions (diagnostic and treatment suggestions) and navigate their professional network. Built Google Colab, LLaMa and Mistral.

NLPLLMsHuggingFaceColabRAGPythonPretrained Models
P02

Instance Segmentation (Mask R-CNN) for Study Space Detection

Fine-tuning the Mask R-CNN deep learning model to detect students and objects in study spaces around the University of New Haven Campus.

Deep LearningHyperparameter TuningFine-tuningModelingNeural NetworksPythonComputer VisionAnnotationInstance SegmentationPyTorch
Code
P03

networker-ai

In development: An AI-powered tool that helps users better understand and navigate their professional network. Built using Google AI Studio, Antigravity, React, Neo4j, Supabase, and Vercel.

ReactAIWeb DevelopmentSupabaseFlaskVercelGitLovableProduct DevelopmentProduct ManagementEntrepreneurship
P04

HangOuts - Check In and Hang Out!

A web and mobile application (in beta on Google Play Store) that allows users to check in at locations in town and see who will be joining in. This app is part of a passion project, aiming to help people connect and socialize in person. The app is built using Supabase, Flask, Vercel, and Python.

PythonApp DevelopmentWeb DevelopmentSupabaseFlaskVercelGit
Code
P05

Urban Crash Risk Radar

A web application that allows users to view the predicted risk of car crashes in a given city. The risk score is calculated using an XGBoost machine learning model trained on 8 years of national crash data. Achieved a grade of A+.

PythonMachine LearningWeb DevelopmentData VisualizationAWSXGBoostData WranglingData Engineering
P06

PlaylistFit: A Spotify-based Music Recommendation System

Project submitted for the course DCSI 6003: Machine Learning, using Spotify data to train a music recommendation system. Achieved a grade of 100%.

PythonMachine LearningData VisualizationFeature EngineeringModel Evaluation