Machine Learning Data Scientist at Apple, working on the next generation of AI intelligence for Apple devices. Previously with Adobe Research, Intuit, and Morgan Stanley. Columbia University, M.S. Data Science.
I build evaluation infrastructure for production AI — the layer between a frontier model and a shipped product.
My work spans Morgan Stanley, Columbia's AC4 Lab, Intuit, and Adobe Research — building RAG systems, multi-agent frameworks, and multimodal MLLM-as-a-Judge pipelines. Today at Apple, I design evaluation systems for Visual Intelligence: scalable, statistically rigorous, and grounded in real-world behavior.
Less interested in "can the model do it?" than in "how do we know it did?"
A mix of agentic systems, multimodal research, and practical ML tools — built outside of work, shared in public.
A fully agentic RAG system over a movie knowledge graph, exposed through an MCP server. Combines semantic retrieval, graph traversal, and tool-using agents to answer multi-hop cinematic queries with verifiable citations.
A transformer pipeline that adapts generated renditions to user context using few-shot prompting and retrieval-augmented personalization. Evaluated against human preferences across 10+ coherence and faithfulness factors.
A browser extension analyzing YouTube watch history for peace-vs-contempt signals. RoBERTa fine-tuned on GoEmotions, DeepSeek-R1 reasoning traces, AWS (S3/EC2) backend, deployed as part of Columbia's AC4 peace research.
An agentic trip planner that reasons over real transit APIs, traffic, weather, and user preferences to produce multi-modal routes with natural-language explanations and counterfactual alternatives.
Co-authored. Proposes an AI-driven framework for quantifying and promoting peace outcomes — combining NLP-based emotion modeling with decision-support tooling.
Read on arXiv →A deep-learning HTR system for automating student paper grading — end-to-end pipeline covering segmentation, sequence decoding, and evaluation alignment.
Read paper →
Morgan Stanley · 2022
Morgan Stanley Technology Analyst Program · 2022
Microsoft Certified
Morgan Stanley Intern Training
Katalyst · equity-in-tech nonprofit
Community of Python educators & learners