Led a team in developing a full-stack platform using Django (Python) as the web server, SQL Server for database management, and Vue.js as the front-end framework. Deployed the platform in Docker managed by Kubernetes with Helm Charts, with Nginx acting as a reverse proxy. Supported the platform and served more than 1000 users across 18 departments.

Designed and built cross-platform Productivity tools within Django to manage development and further requests.

Developed REST APIs using Django REST Framework to integrate with other platforms.

Reduced ticket solving time by 40% through usage of test-driven development (TDD) strategies with unittest and pytest.

Reduced software cost by $100,000 annually comparing to buy vendor products.

Improved performance of operation systems to deal with high-concurrent transactions by implemented a task queuing system using RabbitMQ, reducing API response time from 5-10 seconds to 1 second and enabled system to process hundreds of transactions.

Optimized SQL queries in MySQL InnoDB with indexing. Accelerated overall querying speed by over 70%.

Designed and implemented a real-time ETL pipeline for transaction and payment data using Kafka for data ingestion, Spark Streaming for processing, and Hadoop and MySQL for long-term storage, Hive for short. Orchestrated workflows with Airflow to manage task dependencies and ensure data pipeline reliability. Built visualizations with Power BI to support business decision-making.

Programmed PySpark as the ETL tool, and Power BI for data visualization. Deployed Airflow and MySQL on a Kubernetes cluster (GKE) in GCP using Terraform for infrastructure provisioning and Helm for application deployment.

Implemented MongoDB to manage semi-structured transaction data storage with sharding and replication strategies for availability.

Designed and programmed a full-stack web platform using Django (Python) and MS SQL Server for Operation with more than 30 web pages with React, utilizing AWS Aurora as database engine and S3 for backup, replaced vendor products costed over 50k per year.

Directed successful software life cycle for 3 large-scale projects, increasing average ticket response time for more than 30%.

Developed Python AML algorithms to prevent fraud and illegal transactions. Combined string similarity algorithm, regularized algorithm, cosine similarity algorithm, and other distance algorithms to build a comprehensive, structured, and reliable software product. Optimized algorithms to get 10x faster and 10x more accurate than the old ones.

Led data modeling to re-construct database for 2 systems, with better table relationships, data structure and indexing strategies and partitioning,

Led ETL program includes data cleaning, normalization, feature engineering and so on in Microsoft SQL Server database with SSMS.

Developed SQL queries and stored procedures in SSIS to generate 10+ business reports serving 3 departments.

Built a playbook of the online retailing ecosystem, difference from oversea e-commerce ecosystem and advantages of Firework. This helped Firework to attract dozens of oversea DTC companies to start cooperation with Firework.

Been responsible for research on how short videos and live stream flourished in China, where short videos and live stream e-commerce did very good. The research included but not limited in legal, business volume, anchor strategies and market turning points.

Co-implemented an automatic voice interaction module for robots used in retail scenarios using Python

Realized the module which incorporated ASR(Deepspeech2), NLP(NLTK) and speech generation capabilities, achieved 92% accuracy per human judgement. Our module was helping two of the biggest snacks retailer in China to realize no-human customer service.

Participated in two tender submissions and maintained communication between clients and developers afterwards

Implemented a VPN network to transmit data securely in public internet through a tunnel transport