Bessie Zhang

I’m currently pursuing a B.S. in Mathematical & Computational Science and an M.S. in Computer Science - AI at Stanford University. I have extensive experience in working with both technical and business-oriented teams, allowing me to easily bridge all key stakeholders. At Merico, I have led a team to fully build and implement a financial model that predicted cost and returns. As a student PM at Stanford, my team’s MVP received high compliments from Christina Wodtke, a Stanford professor that specializes in human management. I'm excited to join a collaborative team and provide meaningful contributions of my own.

I have worked in engineering and analyst roles in different fields of finance, including JP Morgan & Chase and Merico Inc. Currently, I am working on the following two independent Machine Learning projects:
1. DeepRhymes: End-to-end Conditional Rap Lyrics Generation 

The main goal of this project is to build a music lyrics generation model that is conditioned on human inputs. Lyric generation models today use post-hoc methods that incorporate rhymes after a differentiable model generates a set of lyrics. However, Bessie is improving on current lyric generation literature by creating an end-to-end conditioned system that automatically generates both the lyrics and the rhyme scheme concurrently with a Transformer encoder-decoder model.

2. Decision Marking under Uncertainty: A gig economy worker’s travel routes

The main goal of this project is to learn a policy based on a worker’s order data and optimize for their goals, such as total money earned, and time saved. Her model is structured upon the Markov Decision Process, allowing for the account of the Present and Future Value of rewards, and thus trains the model on a sequential decision-making scheme.

B.S. Mathematical and Computational Science and Minor in Economics
M.S. Computer Science