All Projects

The Automatic multi-agent Advertisement Systems involving ARIA (Autonomous Revenue Intelligence Agent) is an AI-powered advertising optimization system that autonomously manages multi-platform ad campaigns through continuous experimentation, learning, and adaptation. The system uses LangGraph orchestration to coordinate specialized agents that observe signals, generate strategies, create content, optimize audiences, allocate budgets, execute campaigns, and evaluate results.

NEURORA is an iPad-first therapeutic application that uses mandala drawing and Apple Pencil telemetry to monitor motor and cognitive symptoms for early detection and progression tracking of Parkinson's disease and Alzheimer's disease, providing gentle AI-powered insights for patients, caregivers, and clinicians.

A biotech research evidence mapping platform that automatically discovers, analyzes, and visualizes scientific knowledge from multiple sources including peer-reviewed papers, datasets, and social signals.

VaultCare is a hackathon-ready MVP that demonstrates how to build a voice-based medical appointment scheduler where Protected Health Information (PHI) never enters the LLM. The system treats all LLM reasoning as untrusted compute and proves PHI isolation through comprehensive compliance replay

CHRONOS is an open-source multi-agent collaboration platform that enables AI agents to discover each other by skills, share structured context safely, propose actions, debate decisions, and resolve conflicts through enforceable AI Court rulings

AI-powered emergency response system that analyzes San Francisco's real-time crisis data using natural language queries.

BuildBench AI's mission is to structurally lower the cost of housing by giving small and mid-size builders the same data and decision tools as the biggest developers—so every new home is designed and built at the lowest safe cost, not the highest guess.

PatchPilot is an autonomous debugging engineer that runs entirely inside an E2B sandbox. It reads your bug description or stack trace, analyzes your codebase, searches Exa for real-world fixes, generates a patch, validates it by running your tests, and finally opens a production-ready Pull Request through GitHub MCP, everything is all automatically!

AutoLabDrive is a self-evolving AI co-pilot that analyzes real driving logs, detects scenarios, retrieves and critiques AV research, and overlays the best methods directly on video. Competing agent labs learn from each event, giving engineers a live, ever-improving research engine that reveals what truly works for their data.

Energetic Goose is an on-chain treasury and liquidity platform that tokenizes energy invoices into digital assets, uses AI for automated underwriting, and enables instant global payments and yield optimization through blockchain-based stablecoin finance.

Durin is a global risk-hedging and underwriting platform that connects users’ verified financial data worldwide to build U.S.-equivalent credit profiles, enabling cross-border credit evaluation and compliance-driven underwriting for digital banks.

SafeRun transforms traditional cybersecurity dashboards into living, story-driven experiences. It visualizes real-time threats through animated AI guardians — from calm cats to defensive mecha dogs — blending game design with serious security analytics. By turning complex data into intuitive motion and narrative, SafeRun makes cyber defense both clear and captivating.

Memora lets you ask any life-admin question by voice and get a concise answer plus the exact proof (PDFs, photos, notes). It highlights the supporting snippet, shows evidence thumbnails, and can set follow-ups (e.g., reminders before expiration). All data stays private by default.

Built a functional Chrome Extension that can offer real values to those learning a new language and those with limited general reading proficiency.

Cloned ChatGPT using full-stack MERN framework that can summarize, rephrase, and translate the context.

Led a team of 5 to build Xref, which is a full-stack cross reference platform that helps professors to track students easily, it allows uploading CSV files, manually adding student's information, and searching student based on information.

Developed a simple note full-stack web application using Django and React framework.

This is my internship report at Vivo, where I was in a Software Engineer position and built the platform for camera's automation testing and implemented several carema algorithms.

Predicted the survival factors(passanger id, name, cabin, fare, etc.) of Titanic with Neural Network.

Found the best tuning parameter values through cross-validation and display these values and make prediction on the testing data, report the RMSE and the Coefficient of Determination R2

Split the data randomly into training (75%) and testing (25%), first build the best random forest to predict Spam e-mails using the training data, then use the out-of-bag (OOB) data to measure its performance, and then use this random forest model to predict whether each e-mail in the testing data is Spam or not.

Used all the other 8 variables to predict the binary dependent variable 'diabetes' telling us whether the subject is diabetic or not (factor with 2 levels: neg and pos). You will split the data into 80% training and 20% testing.

Built a full-stack strategy game that allows multiple players play together.

Built a full-stack web application called Battleship Game, which allows two players and one server. Players can move, attack, and perform sonar scan.

Built a traveling app that tested on IOS The advantage of this application is that it can give you all the recommendations for Restaurants, Attractions, and Hotels separately within one search each. (Search based on Google Map API)

This is a prototype built in EST106 to solve the problem of ice cream is easy to melt. This idea won the most vote at EST106 class.

This is a Business Analysis Project for drawing the protrait of Mobike.