The instructors

The standard

One requirement sits above all others: the person at the front of the room does this work for a living, right now, in production. Course material comes out of current projects — kernel trees being patched this quarter, boards being brought up this month, systems being debugged this week — not from a deck written years ago and recycled.

Where that experience comes from

The bench's background spans kernel and systems engineering at Wind River (Linux kernel architecture on the eLxr project), Qualcomm (kernel and embedded at handset scale), Amazon Robotics (systems where latency and reliability are the product), Graphcore (AI accelerator systems) and Spire (embedded systems that had to work, in orbit, with no console cable).

Current production work includes kernel CVE triage and backporting across multi-branch release trees, board support packages and PREEMPT_RT engineering — the same material as the security, BSP and real-time tracks. Research work in FPGA acceleration and neuromorphic computing feeds the FPGA and AI-compute tracks.

Why practitioner-taught matters

Vendor training is well produced and taught by professional instructors working from a script. Bootcamps stay at the framework layer. Very few people who can do kernel-level work also teach it, which is the entire reason this academy exists.

In practice it means the answer to "what happens if…" is usually a real story rather than a slide, the labs break in ways that actually happen, and when your question drifts off the syllabus the answer comes from experience rather than from a trainer's manual.