Projects

Overview

A fleet inspection platform designed with AY Logistics to centralize truck inspection evidence, vehicle issue review, maintenance follow-ups, and operational reporting through a web dashboard.

Problem

Truck inspection workflows often rely on scattered photos, manual notes, messages, and disconnected maintenance records. This makes it difficult for operations teams to review vehicle condition, track unresolved issues, and follow up on maintenance work consistently.

What I Built

• Designed dashboard flows for fleet inspection status.
• Built reusable React UI components for cards, forms, status views, and review screens.
• Designed backend API workflows for image ingestion, issue tagging, damage evidence storage, inspection history, and maintenance follow-up tracking.
• Planned relational data models for vehicles, inspections, issues, image evidence, and maintenance records.
• Structured the product workflow to turn raw inspection evidence into reviewable operational data.

Technical Breakdown

• React frontend dashboard.
• Python/FastAPI backend API design.
• REST API workflows.
• PostgreSQL-style relational data modeling.
• Image ingestion, issue tagging, damage evidence storage, inspection history, and maintenance follow-up tracking.

Overview

An AI-powered platform that matches underserved founders with relevant grant opportunities and gives them a structured pipeline for tracking funding applications.

Problem

Founders often spend too much time searching across fragmented grant databases, eligibility pages, and funding announcements. The process is difficult to personalize, hard to track, and especially painful for underserved founders with limited support.

What I Built

• Built an AI-driven matching workflow using Retrieval-Augmented Generation concepts.
• Designed founder-to-opportunity matching logic based on business profile, eligibility, sector, and funding fit.
• Created grant pipeline tracking workflows for organizing opportunities, application status, and next steps.
• Developed product strategy for a B2B2C model targeting local government and CDFI use cases.
• Designed a clear user workflow for discovering, saving, and tracking grant opportunities.

Technical Breakdown

• Retrieval-Augmented Generation and AI-driven opportunity matching.
• Founder profile matching and grant pipeline tracking workflows.
• B2B2C product strategy for local government and CDFI use cases.

Overview

A full-stack retail platform for managing products, orders, inventory, payments, and business analytics through a modern web dashboard.

Problem

Retail teams need more than a storefront. They need visibility into inventory levels, sales activity, customer behavior, order status, and operational trends. Without a centralized dashboard, business decisions become reactive, manual, and fragmented.

What I Built

• Built a full-stack web application with product, order, inventory, customer, and payment workflows.
• Engineered backend APIs for business logic, authentication, reporting, and data access.
• Created frontend dashboard views for inventory, sales activity, and operational insights.
• Designed database schemas for products, users, orders, payments, and reporting data.
• Implemented Redis-backed performance patterns for faster data access where appropriate.
• Structured payment workflow integration using Stripe.

Technical Breakdown

• Full-stack web application architecture.
• REST API backend.
• Product and inventory management.
• Order and payment workflows.
• Admin dashboard, reporting views, and database-backed business workflows.
• Authentication and authorization workflows.

Overview

A machine learning model that predicts touchdown likelihood using football game and player data. The project applies data preprocessing, feature engineering, model training, and evaluation to a sports analytics problem.

Problem

Football outcomes are influenced by many variables, including field position, player usage, team tendencies, game context, and historical performance. The goal was to turn raw football data into predictive features that estimate touchdown probability.

What I Built

• Cleaned and prepared football datasets for machine learning workflows.
• Engineered features from player, team, and game-context data.
• Trained classification models to estimate touchdown likelihood.
• Evaluated model performance using accuracy-focused and error-analysis metrics.
• Built a repeatable Python workflow for experimentation and model improvement.

Technical Breakdown

• Data cleaning and feature engineering.
• Classification modeling.
• Model evaluation and error analysis.
• Repeatable notebook or script-based ML workflow.

Overview

A Java-based business management system for handling inventory, invoices, customers, and reporting workflows.

Problem

Small business operations often rely on spreadsheets or manual tracking for inventory and invoice records. This creates errors, duplicate work, and limited visibility into stock levels and customer transactions.

What I Built

• Built a Java application for managing inventory, invoices, customers, and transaction records.
• Designed a MySQL database schema for storing products, invoices, customer records, and reporting data.
• Implemented CRUD workflows for adding, editing, searching, and deleting business records.
• Created reporting views for reviewing inventory and invoice activity.
• Built user interface workflows for common business operations.

Technical Breakdown

• Java application structure.
• MySQL relational database.
• JDBC data access.
• CRUD workflows for inventory, invoice, customer, and reporting records.