Factor Model

Rewrite the factor model algorithm which reduces 5,000+ customer models’ scheduled inference time by over 90%, with weekly 500+ hours less computation time on AWS EC2. The algorithm uses numpy, Clickhouse, and PostgreSQL to efficiently compute economic factor values for every publicly listed security and ETF (20,000+ securities) each day.

Portfolio Optimization Algorithms

Developed the investment style matching feature facing 150+ institutional clients using Python, gRPC, and protobuf. The feature analyzes client’s portfolios and reports the fitness of their selected investment style.

Added features in the Boosted.ai trading algorithm with GraphQL to optimize daily stock selection for all the company’s clients. The added features expand the algorithm’s capabilities to construct portfolios that align closer with the client’s needs.

Macro Commentary AI

Developed the AI commentary features facing 150+ institutional clients which use the power of large language models (LLMs) to create textual analysis on the clients' portfolios against various macro topics.