Jessica Foo

Technical Lead for GovTech's Responsible AI (AI Practice) team - Driving research and experimentation on tools and practices for safe and responsible AI across govt

Video Analytics, Data Science and AI Division • Created unit and integration tests to improve continuous integration and deployment practices for computer vision platforms (PyTest, Ward, Jenkins) • Deployed object detection models (PyTorch) on Kubernetes for integration into cloud-native video analytics microservices • Full-stack development with experience in ReactJS, Typescript, Spring Boot, GraphQL, NoSQL, Docker, Serverless - created front-end features, refactored API microservices and deployed AWS resources for video analytics platform • Developed few-shot image classification solutions for ongoing trials with limited data as proofs-of-concept • Leading engineering effort to create React Native mobile app for early dementia detection • Working with policy makers to develop metrics and tools for AI explainability and fairness

Data Analytics Unit, Economics Division • Used predictive algorithms, NLP techniques and network analytics for a variety of projects to aid policy formulation • Tested machine learning causal inference (e.g. generalised random forest, Bayesian networks) and "explainable machine learning" (e.g., partial dependence plots, local interpretable model-agnostic explanations) on policy evaluation research

Global Alternative Premia Team, in the 2nd largest quant hedge fund in the US (by AUM, in 2018) • Applied statistical methods to analyze and replicate style timing academic research and tested multi-strategy trading algorithm based on timing signals

• Cleaned data and analyzed panel set of innovation inputs and outputs, applying fixed effects regression modeling, principal components analysis and basic machine learning methods

• Developed a leasing model with sensitivity and scenario analyses for a redeveloped West Philadelphia church building to include relevant social services programming for the community and sustainable revenue model for the building

• Part of a pro bono consulting team of four that formulated donor outreach strategy for US and international governmental, philanthropic, and corporate donors and researched crowd funding best practices

Managed social media pages (acquiring more than 3,000 likes in 2 months) and piloted inaugural MP3 Experiment in collaboration with ImprovEverywhere, with more than 700 people in attendance

Organized and managed publicity events, projects and client relationships with corporate and educational partners and assisted in creating content for 2 apps and financial education activity book

Research on computer vision and large multimodal models Advisor: Professor Mike Shou

CS221 AI Principles and Techniques, CS231N CNNs for Visual Recognition, CS224N NLP with Deep Learning, CS238 Decision Making under Uncertainty

Relevant Coursework: Statistical Theory Core; Applied Statistics Core; Nonparametric Inference; Machine Learning; Pattern Recognition; Planning, Learning and Estimation for Robotics and Artificial Intelligence Projects: Spam Filtering with Generalized Additive Models, Analysis of Outcomes of Lymphedema Surgery, Using LDA Topic Model to Analyze Text Messages, Predicting and Tracking Robot Movement with EKF SLAM, Tracking Pendulum Movement with an Unscented Kalman Filter Thesis: Macroeconomics and FinTech - Uncovering Latent Macroeconomic Effects on Peer-to-Peer Lending https://arxiv.org/abs/1710.11283

Relevant Coursework: Data Analytics and Statistical Computing, Applied Probability Modeling in Marketing, Mathematical Modeling in Finance, Macro-Modeling in Economics, Advanced Econometrics: Time Series Econometrics, Advanced Econometrics: Microeconometrics, Game Theory, Linear Algebra