I'm an AI/ML engineer focused on agentic systems, LLM orchestration, and applied computer vision — with full-stack engineering as the delivery layer that turns those systems into shipped products. Most of my recent work sits at the boundary between an autonomous agent and the real world it's allowed to act in: reviewing what an agent proposes before it runs, tracking what a model actually sees, and making sure the pipeline underneath stays honest at scale.
Built Sentinel, a three-stage guardrail agent for LLM-powered coding assistants. Sentinel sits between an LLM agent and its execution environment, reviewing every proposed action before it runs — integrating with Claude Code, Cursor, and CodeX.
Automated web-based operations for internal platforms using Python & Flask, and ran performance testing that improved efficiency and software stability at an enterprise scale.
Developed cross-platform web and mobile applications using web technologies and Android frameworks, designing user-friendly, responsive interfaces in close collaboration with peers.
Developed full-stack web applications using React.js & Flask, and implemented SEO strategies that increased web traffic by 15%.
Local-first multi-agent task orchestrator with real-time WebSocket token streaming, permission gating, and empirical action verification.
Real-time sign language gesture detection built on YOLOv8, tuned for edge deployment.
General-purpose object detection pipeline with DeepSORT multi-target tracking for high-throughput video.
Security and surveillance anomaly detection system built on YOLOv11 for real-time alerting.
Signal-processing pipeline for extracting and classifying features from ECG data in real time.
Career management microservices suite — gateway, worker, credential vault, web client, and browser extension.
Real-time group expense tracking and splitting with graph debt simplification.
Booking platform for vehicle rentals with Stripe payments and optimistic locking against double-booking.
GPT-4-powered customer service bot for Split Money app support with multi-turn state.
AI-driven voice capture and processing pipeline built end-to-end in Python.
Local-first multi-agent task orchestrator with real-time WebSocket token streaming, permission gating, and empirical action verification.
Real-time sign language gesture detection built on YOLOv8, tuned for edge deployment.
General-purpose object detection pipeline with DeepSORT multi-target tracking for high-throughput video.
Security and surveillance anomaly detection system built on YOLOv11 for real-time alerting.
Signal-processing pipeline for extracting and classifying features from ECG data in real time.
Career management microservices suite — gateway, worker, credential vault, web client, and browser extension.
Real-time group expense tracking and splitting with graph debt simplification.
Booking platform for vehicle rentals with Stripe payments and optimistic locking against double-booking.
GPT-4-powered customer service bot for Split Money app support with multi-turn state.
Anagentthatcanactisonlyastrustworthyasthesystemwatchingitact.Idon'tshipautonomy—Ishipautonomywithawitness:somethingthatreviews,verifies,andcansayno.
Sentinel and LATO both exist because "the model said so" isn't good enough. Every proposed action gets reviewed against real constraints before it executes.
A guardrail nobody can afford to run gets bypassed. Sub-5ms gateway checks and 35ms end-to-end vision pipelines aren't vanity metrics — they're what makes safety usable.
A model is not a product. I build the FastAPI gateway, the RBAC layer, and the React surface around it — because the delivery layer is where trust is actually earned.
Open to conversations about agentic systems, Model Context Protocol (MCP) infrastructure, computer vision pipelines, or high-impact AI/ML engineering roles.