I'm an AI / Edge Systems Engineer focused on real-time computer vision on embedded hardware.

Most of my day-to-day work lives in private, production systems (Qualcomm EB5, NVIDIA Jetson Orin Nano, PPE detection pipelines, etc.), so this GitHub is a curated set of clean, self-contained examples inspired by that experience — not a dump of company code.

• 🧠 Computer Vision & Edge AI – real-time inference on embedded devices (Qualcomm EB5 / SNPE, NVIDIA Jetson / TensorRT).
• ⚙️ C++ Systems Engineering – multithreaded, low-latency pipelines; async workers; backpressure and robustness instead of “just works on my machine”.
• 📡 Cloud–Edge Integration – S3-compatible uploads, REST APIs, and metadata flows between edge devices and cloud backends.
• 🎯 Problem fixing – inheriting partially broken systems and making them stable enough for production.

I’m in the process of turning my professional experience into small, focused, open-source examples.
These repos are being built as clean-room implementations (no proprietary code, no company details).

• cpp-async-worker-queue
  A minimal C++11 worker-queue implementation (bounded queue, background thread, graceful shutdown), similar to what I use for alert processing and background tasks in real-time pipelines.

• cpp-s3-uploader-libcurl-openssl
  An S3-compatible uploader written in C++ using libcurl and OpenSSL, including AWS Signature v4 signing. Designed for edge devices that need to push snapshots or logs to object storage.

• embedded-snapshot-coordination-example
  A small simulation of how detection threads can safely request snapshots from capture threads using atomic flags and callbacks, without blocking the video pipeline.

• cpp-rest-client-libcurl-json
  A robust REST client example with JSON handling and UTF-8 sanitation, inspired by the kind of defensive coding needed on embedded systems talking to flaky networks and strict backends.

I’m prioritizing quality over quantity here: each repo is small, readable, and focused on one idea.

Languages

• C++ (multithreading, concurrency, low-level systems)
• Python

Edge / CV stack

• NVIDIA TensorRT, ONNX
• Qualcomm SNPE
• OpenCV
• GStreamer, RTSP
• Hardware-accelerated video decode/encode

Cloud & tooling

• libcurl, OpenSSL
• REST APIs, S3-compatible object storage (e.g. DigitalOcean Spaces)
• Linux, Docker, systemd, Bash
• Microsoft Azure (certified), OpenAI APIs

• I care a lot about stability: no hidden crashes, no silent corrupt data, no “it works when we don’t touch it”.
• I prefer simple, explicit designs over clever hacks.
• I like turning messy, inherited systems into something you can reason about and extend safely.

• 📧 Email: dp.martinezb@gmail.com
• 🌎 Location: Chile
  🔗 LinkedIn: linkedin.com/in/dpmartinezb

If you want to talk about embedded CV systems, edge AI, or migrating pipelines from weird hardware, feel free to reach out 🙂