What

Accenture challenged our team to harness data science in solving a real business question: "Where should the next great NYC coffee shop be opened?" We approached this with a blend of geospatial analysis, customer sentiment mining, and ML classification.

How

Data Collection & Feature Engineering: We gathered census data and Yelp reviews to identify key features (foot traffic, income levels, review ratings, etc.) influencing coffee shop success. Model Training: I trained a Random Forest classifier that achieved 84% accuracy in predicting optimal neighborhoods based on location and sentiment indicators. NLP-Driven Insights: I used NLP tools like Hugging Face Transformers and NLTK to extract common themes from Yelp reviews (ambience, drink quality, customer service) and translated that into actionable recommendations.

Impact

Provided data-backed location strategy recommendations to Accenture stakeholders. Demonstrated the power of combining machine learning and qualitative review mining to inform business decisions. Strengthened our team's presentation skills through stakeholder-facing demos and Q&A.

My Role

I played a part in the data collection and feature engineering process, as well as model training and evaluation.

Tech Stack

What

The challenge: Young adults often lack the tools or knowledge to make informed long-term financial decisions. Our team built LFP, a responsive financial planning web app designed to demystify budgeting, retirement planning, and life-event forecasting.

How

Frontend & UI: I led frontend development using React, translating wireframes into clean, responsive UI components. I also designed in Figma to ensure alignment between the design and engineering team. Backend & Database: I collaborated on the backend API using Node.js and Express, setting up user data flow and integrating with MongoDB for persistent user scenarios. Team Coordination: I facilitated weekly stand-ups, managed sprint boards, and contributed to feature prioritization — ensuring the app launched ahead of schedule.

Impact

Delivered an MVP that was well-received by users for its clarity and usability. Accelerated financial awareness for non-technical users through intuitive simulations. Developed critical skills in fullstack collaboration and user-centered design.

My Role

Company: Break Through Tech Product Studio Tech Used: JavaScript, React, Node.js, MongoDB, Figma

Tech Stack

What

BuzzMania is a web application inspired by Cookie Clicker, challenging users to click and accumulate resources in a fun and addictive gameplay loop.

How

Frontend Development: Built interactive game components with React and styled UI using CSS and HTML. Game Logic: Implemented click-based scoring system, dynamic level-ups, and resource management. State Management: Used JavaScript to handle state persistence and real-time user interactions.

Impact

Created a fully functional, responsive web game demonstrating fullstack development skills. Users can engage in addictive gameplay while exploring React and JavaScript capabilities.

My Role

Company: Personal Project Tech Used: React, JavaScript, HTML, CSS

Tech Stack

What

A Python program simulating the historic Enigma Machine, featuring rotors, a plugboard, and a reflector to encode and decode messages securely.

How

Algorithm Implementation: Developed rotor and plugboard mechanisms following Enigma encryption logic. Simulation: Built interactive CLI allowing users to encode and decode messages. Testing: Verified correctness of encryption by comparing input/output across multiple scenarios.

Impact

Replicated a historically significant encryption machine. Enhanced understanding of cryptography concepts and Python programming.

My Role

Company: Personal Project Tech Used: Python

Tech Stack

What

Phreddit is a fullstack web application inspired by Reddit, enabling users to create communities, post content, and engage with other users.

How

Frontend Development: Used React to build dynamic, responsive user interfaces. Backend Development: Developed RESTful APIs with Node.js and Express for posts, comments, and user authentication. Database: Managed user data and posts using MongoDB, ensuring persistent storage and fast queries.

Impact

Delivered a working social platform demonstrating fullstack development capabilities. Enabled user-generated content and community interaction with authentication and moderation features.

My Role

Company: Personal Project Tech Used: React, Node.js, JavaScript

Tech Stack

What

Big Back Voyage is a fun web app like Akinator for food, helping users discover the perfect nearby cuisine by answering a few playful questions.

How

Frontend Development: Built UI with React and styled with CSS to create an engaging, user-friendly experience. Logic & Recommendation: Implemented question-answer logic to narrow down food preferences and suggest nearby restaurants. Data Handling: Integrated external APIs for location and cuisine data.

Impact

Provided a playful, interactive food recommendation experience. Demonstrated integration of frontend, backend logic, and external APIs in a fullstack project.

My Role

Company: Personal Project Tech Used: React, CSS, JavaScript

Tech Stack

What

A chat application allowing users to send messages with a custom reliable transport protocol built on top of UDP.

How

Protocol Implementation: Designed a custom reliable transport protocol on top of UDP. Message Handling: Developed Python-based client-server communication with TCP-like reliability. Testing: Conducted extensive message transfer tests to ensure reliability and error handling.

Impact

Demonstrated understanding of networking concepts, TCP/UDP protocols, and Python implementation. Built a fully functioning chat system with message reliability.

My Role

Company: Personal Project Tech Used: Python

Tech Stack

What

A multi-agent conversational AI built for Okada & Company, transforming NYC commercial property recommendations into a cinematic AI crime-syndicate style experience.

How

Frontend: Built a responsive Next.js UI for chat interaction. Backend: Developed Flask APIs to handle RAG-based queries and maintain conversation state. Agent Personas: Created multiple mafia-style AI agents with distinct personalities for engaging, informative conversations. Data Analysis: Integrated financial, spatial, and QoL metrics to generate property recommendations.

Impact

Delivered a unique AI-powered property recommendation system. Demonstrated fullstack development, AI integration, and multi-agent conversational design. Enhanced user engagement through entertaining and informative agent personas.

My Role

Company: Okada & Company Tech Used: Next.js, Flask, Tailwind, Python, OpenAI

Tech Stack

What

Matcha Bell is a Discord bot designed to monitor popular matcha vendors and send real-time alerts when products are back in stock. With ceremonial-grade matcha selling out quickly, this bot helps matcha enthusiasts secure their favorite products without constantly refreshing multiple websites.

How

Web Scraping & Data Monitoring: I implemented scraping routines using Axios and Playwright to detect stock status and price changes across multiple vendor sites, including Japanese and US-based sellers. Discord Integration: Built the bot using Discord.js to send alerts to dedicated channels and allow users to subscribe/unsubscribe to vendor-specific notifications. Automated Scheduling & Reliability: Used cron jobs for scheduled stock checks and added error handling with retries and delays to minimize request failures and prevent server overload. Interactive Features: Enabled role-based subscriptions, ephemeral replies, and interactive buttons so users can easily manage alerts within Discord.

Impact

Reduced the time users spend monitoring matcha vendors manually. Delivered timely, actionable notifications to Discord communities. Demonstrated automation of web scraping, data monitoring, and real-time alerting in a community-driven platform.

My Role

I designed and implemented the end-to-end system, including web scraping, Discord community creation, Discord bot integration, scheduling, and role-based interactive alerts. I also handled error recovery, stock monitoring logic, and testing.

Tech Stack