Development of algorithms for reinforcement learning. Specifically, software implementation of the algorithms and policies described in the paper Batched Multi-armed Bandits Problems, by Zijun Gao, Yanjun Han, Zhimei Ren, Zhengqing Zhou.

Placed at second place in a three day hackathon, held at Google Zurich headquarters. Used Vertex AI, Looker Studio and BigQuery tables for developing a cloud web app for monitoring and doing predictive maintenance of a motor pool.

Developed software for controlling a DJITello drone through facial and human poses recognition, framed by drone camera. Modified a pose estimation library that uses pretrained gpu computer visions models for correctly estimating human faces even in dark conditions.