Apple
Applied AI Software Engineer Intern- Working on applied AI and ML tooling focused on product quality
Applied AIMachine LearningProduct Quality
Jun 2026 — Present
Cupertino, CA
●AI / ML EngineeringStanford, California
CS + MATH · STANFORD37.4275°N · 122.1697°WDean
Dean
Barrow
CS and Math at Stanford. I'm currently building applied AI tools at Apple. Most of my work is machine learning: training models, building the pipelines around them, and figuring out why they behave the way they do.
01Experience
Where I've worked
Roles across AI, ML, and research
Felicis
Venture Fellow- Analyzed different deals and dug into what each team was building
- Organized events for the Stanford campus and for Felicis
VentureDiligenceEventsJan 2026 — May 2026Menlo Park, CAPerplexity
Campus Partner- Ran events and grew the community of students using Perplexity at Stanford
- Carried student feedback back to the team so it actually reached the product
CommunityAIGrowthSep 2025 — Dec 2025Stanford, CAScale AI
Machine Learning Engineering Intern- Built and improved evaluation pipelines that measured how models actually performed
- Focused on model quality, catching the edge cases that quietly degrade a system
Machine LearningEvaluationModel QualityPythonOct 2024 — Aug 2025San Francisco, CAStanford CURIS
Researcher, Empirical Security Research Group- Built security validation tools for ZDNS, Stanford's fast DNS measurement tool
- Measured how DNS behaves across the internet to find where security assumptions break down
ResearchZDNSDNS SecurityGoJun 2024 — Sep 2024Stanford, CATreeHacks
Organizer- Worked on logistics, sponsors, and the hacker experience for a weekend of over a thousand people
- Handled the year of planning that the weekend ran on
CommunityOperationsEventsOct 2023 — May 2024Stanford, CAJane Street
Academy of Math and Programming- Worked through algorithms and probability with people from the trading floor
- Built a lot of the quantitative foundation I still use
AlgorithmsProbabilityMathJul 2023 — Oct 2023New York, NYFermilab
Particle Physics Researcher- Analyzed experimental data to pull real signal out of a lot of noise
- It's where I first started working with messy experimental data at scale
C++Data AnalysisPhysicsMay 2021 — Jun 2023Batavia, IL
02FocusStanford CURIS
Most of my work is machine learning applied to real systems.
Day to day that means training models, building the pipelines that feed them, and digging into why they behave the way they do. I like that you can measure whether something actually works.
The same interest led me to research at Stanford, where I spent a summer with the Empirical Security Research Group measuring how real systems behave in the wild. Model or network protocol, I like building the tools that turn messy behavior into something you can reason about.
03Selected Work
A few things I've built
Each one is open source, links on GitHub
01 / Computer Vision / Full Stack
Savor
Real time food waste intelligence for dining halls.
A full stack platform that watches a dining hall in real time and tells operators exactly what food is being wasted. It detects waste from live camera frames with vision models, dedupes events with structural similarity checks, syncs Stanford menus, and turns everything into dashboards and smart purchasing recommendations.
FastAPIOpenCVVisionReactSQLite
202502 / Machine Learning / Simulation
Penalty Shootout Simulation
Predicting penalty kicks from real World Cup data, then simulating them in 3D.
I took real World Cup shootout data and built models that predict whether a player scores, using features like favored foot, where they aim, and whether it is an elimination kick. The analysis spans Monte Carlo, Bayesian, and Thompson sampling approaches, then feeds a 3D simulation that plays the shot out on the pitch.
Pythonscikit-learnThree.jsBayesian
202503 / ML Infrastructure
ml-infra-vision
A computer vision pipeline built around data quality.
An ML infrastructure pipeline for computer vision that treats data quality as a first class concern. It handles edge case detection and validation so that models train on clean, representative inputs, the same instinct I brought to model work at Scale.
PythonComputer VisionData Quality
202504 / Systems
HeapScope
A miniature memory leak analyzer for macOS on Apple Silicon.
Point it at a running process and it answers two questions: which allocations are unreachable from any root, and which of those form retain cycles. It attaches over the Mach interface, snapshots every malloc zone, scans thread stacks and data segments for roots, then runs BFS and Tarjan to surface leaks. Ships with a React and D3 force graph viewer that renders the whole heap.
SwiftMachReactD3TypeScript
2025