• Collaborated closely with the co-founders to define the product roadmap, prioritize features, and architected the frontend, backend, and machine learning AI systems for B2C SaaS for research universities to leverage in scientific studies• Designed AI food recommender systems, enabling accurate meal recommendations from combinations of 2M+ USDA foods• Designed and implemented Azure OpenAI transformer-based LLM to help chatbot interactions with users assist with food/weight logging, recommendations, nutrition information, and comparisons between foods/meals• Aggregated, cleaned, and processed 5 USDA food databases into datasets used for training LLM, AI, and ML functionalities• Responsible for the deployment and maintenance of the AWS production server for ML and backend SQL services

• Implemented DGLM to model and predict home insurance premiums for dwellings across the Northeast region of the United States• Lead migration of legacy SAS code and ML models to updated, modernized, cloud-based Python architecture• Implemented ML pipeline to clean, process, and train models on a dataset of 90M+ samples with 1.2k+ dimensions• Lead research of modernizing property premium pricing techniques using GBM-based ensemble models• Created and maintained research AWS EC2 instances to optimize pricing while maximizing computational ability for models

• Presented and helped design two, hour and a half sessions per week to teach the basics paradigms and techniques• Taught data structures, basic algorithms, recursion, and object-oriented programming to undergraduate students

• Chatbot - Design and implementation of two baseline models, a rule-based approach and information retrieval (IR)-based approach, then an encoder-decoder model to create generative automatic chatbot responses• USDA Model Solutions Competition - Identify new approaches to linking Department of Agriculture (USDA) Economic Research Service (ERS)’s Food and Nutrient Database for Dietary Studies (FNDDS) to Information Resources, Inc. (IRI) retail store data in producing the Purchase to Plate Crosswalk (PPC).

• Created a bidirectional LSTM neural network, using TensorFlow, to predict insurance claim severity• Created two baseline models, Logistic Regression and Naive Bayes (using SciKit-Learn), to compare advanced Neural Network• Used NLP to analyze 100k+ claims with 100+ documents per claim to enhance efficiency of claim professionals• Used AWS SageMaker & EC2 environment to implement, modify, and translate between dev and prod environments

• Analyzed and processed experimental data to find relationships between novel compounds and effects• Migrated legacy, on-premise Oracle Database to new AWS Postgres Aurora relational database• Created new data pipeline scripts, using Numpy and SKLearn, in Python to replace legacy Oracle Pipeline Pilot.

Programmed proof-of-concept visual monitoring system, using Python and Raspberry Pi with a camera module, to provide the Scientists a 24/7 visual imaging to allow scientists to constantly monitor test subjects' health and condition.Started prototyping for updating current, legacy inventory management systems to a web application using Django and AWS.

Professional pet care, specializing in small animals. Services include: pet sitting, dog walking, animal grooming and overall pet wellness.

Managed pizza/pasta orders and delivery service throughout the Lamorinda/Walnut Creek area.