Jessica Yung

Research Engineer

2022

Main projects: - AlphaProof (AlphaZero with LLMs for mathematical reasoning) - SynthID (image watermarking) - Video compression with MuZero

Venture Scout

2018 - 2022

Cofounder and Chief Science Officer

2021 - 2021

Helping auto-patch vulnerabilities in code in seconds rather than days or weeks.

AI Resident

2019 - 2021

Researched transfer learning for computer vision. Published state-of-the-art BigTransfer (BiT) model at ECCV, open-sourced code, wrote TensorFlow blog post. Co-author of MLP-Mixer (NeurIPS). Created and open-sourced synthetic dataset SI-SCORE to analyse robustness of image models in a fine-grained way. Co-first author of robustness study at CVPR, workshop paper at ICLR. Worked on developing more recommender systems that are more user-experience-centric instead of just maximising profits and engagement (and so often maximising for things like anger and polarisation in the process). Co-author of NLP work on recommender systems. Gave TensorFlow tutorial workshop at Perspektywy Women in Tech conference 2020.

AI Engineering Consultant

2018 - 2018

Machine Learning, Python, PyTorch, AWS, Google Cloud Platform - Researched augmenting models with external memory. - Contributed to open-source implementation of Relational RNNs (Santoro et. al., 2018). - Researched methods for time series prediction and unsupervised learning.

Founder and developer

2016 - 2018

Debating motions database with more than 2.5 million pageviews. Easily find a random debating motion or look up previous motions to prepare for upcoming tournaments. The site is maintained with the help of beyond incredible volunteers from the debating community. (I'm still doing this! I will push a new Flask version in Dec 2021 with motions tagged with categories using Transformer-based models and advanced filtering for search and random motions.)

Quantitative Researcher Intern

2017 - 2017

Machine Learning, Trading and Market Mechanics, Statistics, Python, C++ - Researched data preprocessing and modelling techniques for predicting financial data. - Built machine learning research frameworks in Python to facilitate experiments on high performance computing clusters.

Discover Digital (Insight)

2017 - 2017

- Discussed and presented findings on how to improve mobile operator's profits and increase market share based on company data - Designed new prototype for mobile app with student team - Previously also selected for McKinsey Discover 2016

User Experience Design Intern

2016 - 2016

- Overhauled design: removed unnecessary features and repositioned key elements. - Designed wireframes iteratively and developed interactive prototypes. - EquitySim went on to win a Fintech award from the UK Government and got into 500 Startups.

Institutional Securities Spring Insight Intern

2016 - 2016

Work shadowed Investment Banking Division (IBD), Sales and Trading and Equity Research. Fast-tracked to Assessment Centre.