I am currently a ML engineer at Meta working within the Business AI org generally working on monetizing LLMs.
Currently, I support Meta Business Agent.
Previously in this role, I worked on a Meta support chatbot embedded on Shopify websites and new agentic ad experiences.
And prior to that, I worked in Meta Monetization and WhatsApp.
Outside of work, I enjoy studying Vietnamese, throwing pottery, and playing couch co-op video games with friends and family.
University of Pennsylvania, School of Engineering and Applied Science
Master of Science in Engineering in Computer and Information Science
Thesis: Empathic Conversation Models For Distressful Events
Advised by Professor Lyle Ungar
Developed empathetic LLMs using ParlAI to discuss distressful news by evaluating fine-tuning on curated empathic data, long-term memory, retrieval modules, loss functions, and varying context depths. Studied early LLMs and conversational agents (GPT-2, GPT-3, BlenderBot) across prompt design, context management, subjective human evaluation, and ML eval benchmarks for NLP.
University of Pennsylvania, School of Engineering and Applied Science
Bachelor of Science in Engineering in Computer Science
University of Pennsylvania, The Wharton School
Bachelor of Science in Economics
Concentration in Operations Management
Senior Machine Learning Engineer
- Developed new agentic e-commerece experiences on a variety of Meta surfaces including Meta ads, WhatsApp / Instagram, and a Meta-owned shopify plugin
- Led the development of new agentic capabilities and conversation eval and improvements.
Software Engineer
- Rearchitected and improved how interactive message types are parsed and rendered on WhatsApp Android which now enables more than $1 million in daily revenue for the business messaging org
Software Engineer Intern
- Automated 29% of support tickets by implementing an intelligent root cause analysis tool written in Lua and TypeScript.
- Built a distributed trace visualizer for 1000+ Bloomberg services using Ogma to identify performance bottlenecks.
Teaching Assistant for Applied Machine Learning
- Developed theoretical and applied problem sets across core machine learning topics, including optimization, loss formulations, regularization, and deep neural networks.
- Guided capstone ML projects from conception to deployment, debugging training pipelines and advising on multimodal architectures (e.g., zero-shot vision models for fine-grained bird classification)
Teaching Assistant for Decision Processes
- Supported instruction of students in decision theory, business psychology, and behavioral economics
Chief Information Officer
- Led IT team of 8 managing operations of a $8M AUM credit union which included negotiations with software vendors, migrating core banking systems, and developing in-house fraud detection algorithms
Research Assistant
- Conducted research within the intersection of neuroscience and machine learning within the Kording lab.
- Engineered PyTorch DNN probes to decode intermediary representations of pre-trained VGG architectures, evaluating whether computer vision feature spaces exhibit human-like optical illusion susceptibility
- Presented original findings, “Shared visual illusions between humans and artificial neural networks,” at the 2019 Conference on Cognitive Computational Neuroscience
- Contributor: Nature Communications, “Efficient neural codes naturally emerge through gradient descent learning” (Acknowledged for engineering model fine-tuning pipelines and training infrastructure)
Software Engineer Intern
- Created dev-tools for WDL and Cromwell for high-throughput DNA sequencing, reducing debug-time for biologists which in turn accelerates fundamental research in genomics and cancer
