AI sleep & acoustic systems
On-device intelligence within severe power, comfort, and signal constraints — where the body is the product environment.
Research → product IP
Intelligence and security that survive silicon constraints.
Applied research investigating AI, cybersecurity, and intelligent-hardware problems with a path toward owned capability in models, architectures, and signals that survive real silicon budgets and product threat models.
Partnership
University of Glasgow — PhD-level AI cybersecurity
Research leadership under Nizar Bouguerra — Head of R&D at Axon Labs, PhD candidate in Computing Science at the University of Glasgow, and inventor on Flexxon-assigned patent families covering host–storage authentication and AI-based malware anomaly detection in storage devices. Prior Head of R&D at Flexxon (Singapore) and CTO roles spanning AI cybersecurity product strategy and wearable hardware. BSc (Hons) Cyber Security and Forensic Computing, University of Portsmouth.
Nizar Bouguerra — engineering profileSelected public IP (leadership)
Patents that show the depth of the bench
US 11,610,026 B2
Module and method for authenticating data transfer between a storage device and a host device — Nizar Bouguerra (Flexxon)
IL 290159 B2
Shared artificial intelligence processor detecting malware-caused anomalies in storage devices — Nizar Bouguerra (Flexxon)
US 11,509,995
Artificial intelligence based system and method for generating silence in earbuds — Abhi Para
SG / EP / AU families
Storage authentication and anomaly-detection counterparts across Singapore, Europe, and Australia
Focus areas
Problems with a product path
AI cybersecurity at memory
Intelligent protection closer to hardware and memory — continuing the line of work on storage-side authentication and anomaly detection that cannot be faked at the software layer alone.
On-device AI systems
Models and architectures shaped around real compute, memory, and energy budgets for continuous operation — not cloud-only demos.
Research discipline
How we decide what is worth investigating
01
Product path required
Research is judged by whether it can become architecture in a product — power, memory, latency, and security included.
02
Legal and academic gate
Publications and IP claims remain subject to academic and legal approval. Public claims are deliberate.
03
Silicon-first evaluation
Models and signals are evaluated against real compute, energy, and threat models — not cloud-only demos.
When the problem needs new capability.
Research engagements connect to product architecture — not papers for their own sake.