Implemented goods recall strategies provided by product managers in order to eliminate low correlation results in order to increase CVR and CTR.

Designed and implemented a simple training framework via Jax so that deep learning engineers can implement test models to replace Tensorflow in order to speed up the training process.

Designed and led 3 engineers to integrate basic embedding searching ability into our new RPC frameworks to support all our embedding searching needs which be used by ten teams that altogether serve one million daily operations, with the availability of 99.99%.

Encapsulated Kafka, Redis, and most commonly used clients as plugins and Developed streaming protocol for Tencent RPC framework in Python that provides a united framework for all backend development teams in Tencent.

Designed and implemented APIs as a platform so that customers can manage their data more easily and integrate their business logic with our basic AI abilities.

Developed MQTT client and device shadow management with Golang in Device Management service for intelligent devices, designed, developed, and led the process of device interactions between the cloud side and device SDK, which greatly improved customer usability.

Designed and developed a distributed data-redistribution micro-service system to support upstream data transmission and Redesigned heavy-weighted middleware for private deployment.

Developed a user profile management service with node.js using the Koa framework in order to help engineers to keep track the status of the maintenance status.