Researched and developed innovative technology using a combination of sharding + POS + POW (first in the world) to improve current crypto transaction speed by 3 times. Researched on the application of AI in a blockchain-based payment system, resulting in a 35% improvement in payment processing efficiency.Researched in traditional payment and how to use Web3 technology to improve efficiency and security.Developed technical framework to improve traditional payment gateway with blockchain and crypto.Implemented Web3 payment and AI features on existing FIDOM payment gateway, increasing transaction volume by 200%.

Conducted a comprehensive analysis to identify 43 critical pain points in the Nut Wallet, specifically within the Transaction and Asset Modules, optimizing the wallet for commercial needs. Successfully implemented these enhancements in projects across Memeland and parts of Africa.Developed a bridging solution that integrates Web2 social dynamics with Web3 on-chain token functionalities, leveraging the Reddit API to synchronize Reddit points with a blockchain database on the Polygon network. This enabled a direct 1:1 on-chain conversion of points, supported by detailed technical documentation to ensure robust integration.

Responsible for mining the company’s operations data, analyzing, and presenting data visualization reports for 89 consecutive days.Engaged in the design of data automation requirements and executed code implementation. Independently developed 3 scripting codes. And assisted automatic driving data development requirement analysis and code implementation.Annotated 767-point cloud datasets and 847 image datasets to serve as data sources for algorithms. Identified a semi-automated annotation model, reducing manual labor costs by 10-30%.Summarized various issue documents, produced 16 technical documents for the company, and set up the PlotJuggler data visualization platform to aid in analysis.

Developed a university encyclopedia answering service with an NLP-based chatbot at its core, read user messages through NLU so as to understand the user’s intention, then the rule engine calculates the best response, and finally pass the data back to the web chat service through developing flask web tool. Collected and used more than 1300 questions to train the language model, and by summarization and observation the accuracy of the model reaches over 93%.