{"id":95,"date":"2026-03-23T15:16:03","date_gmt":"2026-03-23T15:16:03","guid":{"rendered":"https:\/\/numidai.com\/?page_id=95"},"modified":"2026-09-05T20:13:03","modified_gmt":"2026-09-05T20:13:03","slug":"about","status":"publish","type":"page","link":"https:\/\/numidai.com\/index.php\/about\/","title":{"rendered":"About"},"content":{"rendered":"\n\n\n
\"NumidAI<\/figure>\n\n\n\n

\u2014 About<\/p>\n\n\n\n

\n\"N.G\"\n
\n

Nadir GHEZALI<\/strong><\/h1>\n

Founder \u00b7 Software Engineer<\/p>\n<\/div>\n<\/div>\n\n\n\n

\n\n\n\n

The project<\/h2>\n\n\n\n

NumidAI is an independent research and development project focused on autonomous wildfire detection at the edge.<\/p>\n\n\n\n

This project is personal. Wildfires cause devastating damage every year across the world \u2014 including in the region my parents are from, where lives have been lost and vast landscapes of olive groves have turned to ash.<\/p>\n\n\n\n

Early detection is crucial in wildfire management. Catching a wildfire in its first minutes can mean the difference between containing a small blaze and losing large areas of vegetation and potentially lives.<\/p>\n\n\n\n

\n
\n\n\"Kabylie,\n<\/a>\n
Kabylie, Algeria – 2020. \u00a9 EPA<\/figcaption>\n<\/figure>\n
\n\n\"Aude,\n<\/a>\n
Aude, France – 2025. \u00a9 Occitanie Vue du Ciel<\/figcaption>\n<\/figure>\n<\/div>\n\n\n\n
\n\n\n\n

Background<\/h2>\n\n\n\n
\n\n

Experience<\/p>\n\n\n

20+ years in embedded systems and real-time software. C++ and Linux specialist. Experience in video processing, edge computing, and software architecture.<\/p>\n\n<\/div><\/div>\n\n\n\n

<\/div>\n\n\n\n
\n\n

Why NumidAI<\/p>\n\n\n

NumidAI is the concrete application of embedded systems expertise to a problem that matters: early wildfire detection.<\/p>\n

I have worked on every single aspect of the system<\/strong> \u2014 from initial specifications and requirements, through data engineering, model training and evaluation, to embedded integration and optimization on edge hardware.<\/p>\n

Data<\/strong> \u2014 Dataset aggregation and curation at scale. Metadata taxonomy design. Hard negative mining. Model-assisted labeling. Stratified splitting with strict data leakage prevention.<\/p>\n

Model<\/strong> \u2014 Recall-first detection strategy. Signal-aware training distribution. Threshold calibration under operational false-positive constraints.<\/p>\n

Embedded software<\/strong> \u2014 Real-time C++ inference pipeline on edge hardware. Multi-threaded architecture with sub-100ms end-to-end latency. Multi-layer detection and semantic filtering.<\/p>\n

Validation<\/strong> \u2014 Benchmarked on FIgLib (HPWREN), the academic reference for camera-based smoke detection, on the published state-of-the-art test split, in strict out-of-distribution: 100 % of fire events detected. Also exercised on hard sites and live streams worldwide (HPWREN, explore.org, Cannon Mountain), from 720p to 1080p, in grazing light and confuser-heavy scenes. Full protocol and result tables available on request. Operator web interface served by the module itself.<\/p>\n

Everything else<\/strong> \u2014 System architecture, deployment strategy, field evaluation protocol, technical documentation, and this website.<\/p>\n

A project built to validate a technical hypothesis on real hardware, using real data \u2014 now benchmarked against the published state of the art, and looking for the field.<\/p>\n\n<\/div><\/div>\n\n\n\n

\n\n\n\n

Open to<\/h2>\n\n\n\n
\n\n