AI/ML Engineer
I build AI-powered systems that work in the real world - from developing and fine-tuning models for domain-specific tasks, to deploying real-time robotics, to shipping intelligent applications into production.
Currently at Araca, I'm building an analytics platform for semiconductor research, starting as a desktop application with interactive dashboards, ML prediction models, and an LLM-powered assistant, with plans to expand into a web-based platform. Previously, I designed AWS pipelines processing 500K+ events daily and built a coding platform serving 300+ students.
My MS thesis at the University of Arizona tackles vision-language-action models for autonomous driving, fine-tuning a 957M parameter model and deploying it in real-time control loops at 4Hz. I've published three research papers spanning autonomous vehicles, drone swarm coordination, and VR teleoperation.
End-to-end machine learning - from data pipelines and model training to deployment and monitoring. Fine-tuning LLMs, building prediction models with proper cross-validation, and integrating AI into production workflows.
Building systems that perceive, decide, and act in the real world. Real-time control loops, sensor fusion, and multi-agent coordination with published research.
From responsive frontends to scalable backend services, REST and GraphQL APIs, relational and NoSQL databases, authentication and authorization, containerized deployments, and CI/CD pipelines. Desktop apps, web platforms, and the infrastructure that connects them.
Embedding intelligence into real products - not chatbot wrappers. Fine-tuned LLM-powered assistants for domain-specific workflows, AI-driven analysis tools, and intelligent automation that actually ships.
Each one built end-to-end. Click to see the full case study.
Analytics platform that replaced 100+ hours of manual spreadsheet work for semiconductor researchers. Automates sensor data processing, interactive analysis, ML-driven predictions, and one-click report generation. Features an AI assistant for intelligent data interpretation.
Vision-language-action model for autonomous vehicle control. Fine-tuned a 957M parameter model using LoRA, designed a complete data collection pipeline, and deployed real-time inference at 4Hz with PID-tuned control.
Immersive third-person VR system for aerial robot control. Built a low-latency WebXR-to-ROS2 bridge on Jetson Orin with real-time SLAM, increasing obstacle clearance by 0.20m without compromising task time.
Experimental validation of decentralized affine transformation for multi-agent aerial systems. Six Crazyflie quadrotors coordinate formation changes through constrained environments using only local neighbor communication - no global planner needed.
Open to opportunities.