- Created and deployed deep learning models with LSTM’s and attention to predict prices for trades of 1M+ municipal bonds, using TensorFlow and PyTorch- Enable users to price up to 100k bonds at once by designing and implementing a robust, scalable, cloud-based serverless backend on Google Cloud Services- Sped up data cleaning and processing by 95%, ensuring data accuracy and reducing processing time- Designed and developed the user interface, including the user authentication systems using Firebase- Reduced latency by 70% and cloud service costs by 90% by pioneering and developing an efficient pipeline for collecting real-time trade data- Built automated testing frameworks using Pytest to monitor the health of the product, reducing server downtime and ensuring the product's reliability

Interested in classification/regression problems with decision-dependent data (e.g., performative prediction, strategic classification) and online optimization (e.g., sequential decision making, resource allocation).- Developed novel algorithms for expected loss minimization for a decision-dependent time-varying data distribution, and implemented these algorithms on real-world parking data using NumPy and scikit-learn- Created a novel surrogate function design technique to maximize competitive ratio for a sequential online resource allocation problem, and implemented this technique using CVXPYDecision-dependent risk minimization: https://arxiv.org/pdf/2204.08281.pdf (published in AAAI) Online optimization and resource allocation: https://arxiv.org/abs/2012.12457 (in submission)

- Handled course logistics, course material, exams, and final grades- Gave lectures and recorded associated lecture videos- Interviewed, hired, and mentored a teaching staff of 30 TAs and tutors- Expanded tutor responsibilities to give them more teaching and content development experience- Developed formative assignments for academic interns to better prepare them for labs and office hours

Project: Best of many worlds: Robust model selection for online supervised learninghttps://arxiv.org/pdf/1805.08562.pdf (published in AISTATS)

CS 61A: Summer 2016, Fall 2016, Summer 2017 (Head), Fall 2017, Spring 2018CS 61B: Spring 2017Head uGSI Duties: managed staff and course logisticsDuties: taught sections, developed course materials, held office hours, graded assignments and examsCS 61A is an intro computer science course on abstraction, functional programming, and OOP (Python)CS 61B is a computer science course on data structures, runtime analysis, and software engineering (Java)

Project: Open-source Automated System for Assembling a High-Density Microwire Neural Recording ArrayPublished in: MARSS, International Conference on Manipulation, Automation and Robotics at Small Scales. July 2016

Software Test Automation Team- Wrote tests and improved current framework to increase testing coverage within a singleapp and across multiple apps for the ShareFile android application

DataPower Software Quality Assurance (SQA) Team- Researching automation technologies to expand DataPower testing environment

Biostatistics BranchProject: Minimizing Systematic Errors in Quantitative High Throughput Screening Data Using Standardization, Background Subtraction, and Non- Parametric RegressionPublished in: The Journal for Experimental Secondary Science (JESS): Ray, M. et al. April 2014, Vol 3, Issue 2; Page 1-5; ISSN#2162-8092Mentors: Dr. Grace Kissling, Dr. Keith Shockley