Software Engineer with 9+ years across Meta and Amazon, focused on large-scale distributed systems, platform security, and experimentation-driven product development. Currently leading security and integrity work for Instagram account recovery — designing policies, running A/B experiments, and prototyping LLM-based remediation. Previously delivered full-stack and cloud infrastructure for Amazon Pharmacy and Amazon Fulfillment.
A compiled, auto-parallelizing systems language built end-to-end — lexer through native LLVM and WebAssembly codegen. Rust-style ownership/borrow checking with no lifetime annotations, an effect system that infers reads/writes and auto-parallelizes independent work, generics via monomorphization, and 10,000+ tests with ASan/LeakSanitizer leak gates on every build.
Two image tools whose compute kernels are written in Kāra and compiled to WebAssembly by its own backend — the browser only decodes and re-encodes. Prism is an image workbench: Lanczos-3 and bilinear resampling, crop, rotate, flip, and brightness/contrast/saturation adjustment. Veil redacts screenshots with solid-bar, pixelate, or blur, stripping EXIF and GPS data on export. Both run entirely on-device.
Surfaces clickbait flags and information-density ratings on YouTube videos, with community averages. Serverless: Firestore via REST (no SDK), MV3 service worker, Google OAuth.
A lightweight browser-action todo popup built on Manifest V3. Drag-and-drop task reordering, light/dark themes and adjustable text size, all persisted locally through chrome.storage — no backend, no dependencies.
Lead security and integrity work on Instagram's account recovery surface. Shipped policies that cut account compromises 40% (~92K/year) and prevented ~20K account takeovers/year. Drove the team's first Python→Hack backend migration and prototyped an LLM-based system to auto-reverse false-positive enforcements (projected 1.8M MAU/week).
Full-stack development on Amazon Pharmacy's customer onboarding. Designed a micro-frontend architecture for independent team deployments, built a reusable CDK library for centralized monitoring, and led migration of a Java Spring service from a shared to a dedicated AWS account.
Built data ingestion infrastructure-as-code for Amazon's fulfillment data platform. Established cross-account ingestion standards via API Gateway and Kinesis, authored reusable CDK components for ETL pipelines, and developed RPA automation on Windows EC2 that saved $500K+/year.
Designed experimental studies and A/B tests for ad products at Facebook and Google. Built Python pipelines on Presto and Hive, used BigQuery for user segmentation, and shipped Tableau dashboards for campaign performance and sentiment trends.
Automated reporting and ETL for the Alumni Association in Python and Tableau, saving 20 hours of analyst work per week. Built dashboards to monitor business KPIs and surface marketing insights.
Built backend logic in Oracle PL/SQL for a T-Mobile CRM application supporting new telecom initiatives. Resolved critical production issues within SLA.