Edge AI & Computer Vision Engineer
I turn AI models into reliable systems running on edge devices — from model integration and performance optimization to containerized deployment, device security, and production operation on NVIDIA Jetson.
I currently work with C++, Python, Linux, Docker, computer vision pipelines, and embedded AI systems. My public work focuses on reproducible engineering: real hardware, measurable results, explicit trade-offs, and documentation that includes what failed as well as what worked.
A practical engineering lab for Edge AI on the NVIDIA Jetson Orin Nano using JetPack 7. It documents reproducible setup, GPU-enabled containers, deployment behavior, benchmarks, architecture decisions, and the operational details behind running AI workloads on constrained devices.
Current focus: Docker and GPU execution on Jetson, native-vs-container measurements, power-mode experiments, and reproducible infrastructure for future inference, video-pipeline, TensorRT, security, and smart-camera chapters.
NVIDIA Jetson · JetPack 7 · Docker · CUDA · FastAPI · Benchmarking · Edge AI
A C++17 benchmarking tool for ONNX Runtime inference, with a Python reference implementation for cross-validation. It reports latency distribution, percentile metrics, throughput, warm-up behavior, and model metadata.
C++17 · ONNX Runtime · CMake · Python · Performance Engineering
An experimental tennis-video analysis pipeline combining a multithreaded C++ application with Python/FastAPI inference services. The project integrates player, ball, and court-keypoint models and processes video frames through bounded queues and shared application components.
Computer Vision · C++ · Python · FastAPI · YOLO · OpenCV
- Computer vision inference and video-processing pipelines
- Edge AI deployment on NVIDIA Jetson
- C++ and Python systems for real-time AI applications
- Docker-based delivery and operational reliability
- Inference benchmarking, latency analysis, and performance optimization
- Secure deployment, device encryption, and model-protection workflows
I am building toward end-to-end ownership of edge AI systems: model → optimization → deployment → device security → operation.
My current public roadmap is centered on expanding edge-ai-lab with:
- ONNX Runtime and TensorRT inference on Jetson
- OpenCV and GStreamer video pipelines
- Reproducible performance benchmarks
- Smart-camera architecture and observability
- Secure Boot, encryption, hardening, and deployment security
- Quantized vision-language models running at the edge
Languages: C++, Python, C, JavaScript, SQL
AI & Vision: Computer Vision, Deep Learning, ONNX Runtime, TensorRT, OpenCV
Edge & Systems: NVIDIA Jetson, Linux, CUDA, Docker, CMake, MQTT
Engineering: Performance optimization, containerized deployment, system architecture, secure delivery
- Intermediate AI Developer at Vision Tech Consulting
- Former R&D Computer Vision Developer at DtLabs
- B.Sc. in Computer Science from the University of São Paulo (USP)
- Top-3 finalist in the ITA Flight Delay Prediction Challenge
I am especially interested in international work and open-source collaboration involving Edge AI, Computer Vision, NVIDIA Jetson, inference optimization, and production AI systems.