Uses GCP, Firebase, Phaser, and React to develop a blockchain based grand strategy game (https://grandconquest.com).Frontend: 1. Create React components with Styled Components to interface with Phaser game engine.2. Leverage Lexical to create custom in game chat system featuring a rich editor functionality.3. Ensuring real-time data accuracy by synchronizing and integrating information from both blockchain and Firebase Firestore, prioritizing freshness from blockchain while efficiently managing partial data retrieval from blockchain to maintain seamless user experience.Backend:1. Implement rolling update strategies to attain eventual consistency between Firestore and blockchain2. Implemented custom mutex locking mechanism to avoid dirty write and wasted read

Developed internal applications for factory use. Tools used include Python, FastAPI, Pandas, OpenCV, Express & React.Highlighted Applications: 1. CCTV Monitoring System: A web app that streams SSE live/playback videos based on the users' selection of camera(s) in the frontend. Additionally, an admin tool for manager to update users' access rights and add/remove cameras' positions.2. Workload Scheduler & Performance Report Generator:Retrieve factory's machinery, production, and client data from ERP's database and generate estimated workload schedule as Excel and PowerPoint for executives.

Built desktop applications to automate mission logistics filing with Python, Azure OCR, OpenCV and VBA.

Transform company's website to a WYSIWYG website with Typesetter CMS for non-technical staff

Conducted research on building real-time edge computing systems using Elixir, Rust, Go and stream processing.Highlighted Presentation:-- EdgeSys: A Decentralized Cloudlet Framework for Edge Computing (HotEdge'20) - Used Elixir to build a framework that run clients' workload and maintain soft real-time requirements. - Supported workload includes Storm API, WASM, and GRPC applications. - Used Multi-agent Path Finding (MAPF) application to demonstrate the framework's fault tolerance.Highlighted Publication:-- Pacer: Automated Feedback-Based Elasticity for Heterogeneous Soft Real-Time Workloads (IEEE UCC'18) - A Python framework to allow application developers to deploy custom feedback-based resource allocation algorithms on the VMs under the same hypervisor.

-Built new Wall Street Journal's online subscription page in a team of nine-Frameworks used include Salesforce, ExpressJS, Ruby on Rails, and MongoDB-Responsible for various backend data processing using python and JS

-Helped The Huffington Post to develop web forum troll detection algorithm base on contents only-Developed custom deep neural network in Java-My partner and I delivered a 70% accuracy result which is similar to the result delivered byThe Huffington Post machine learning team which composed of more than 10 people

- Developed a high speed optical wireless communication link protocol with visible light(LED)- Main responsibilities include writing FPGA modules to implement various modulation schemes such as PAM and PPM for both transmitter and receiver

- Collaborated with MaXentric Technologies to create gateway for communication between devices- Implemented encryption for a secure communication between servers by utilizing VPN and various communication protocol such as XMPP and SIP

- Collaborated with Huffington Post to develop iphone app prototype- Implemented handgesture interface with Objective-C