XENTORYX LABS // R&D EXPERIMENTS

Innovation & Experimental Hardware Hub

Exploratory R&D prototypes, TinyML edge model quantization on ESP32 microcontrollers, high-frequency MQTT brokers, and adaptive low-power drivers designed at Xentoryx Labs.

PrototypeQ1 2026

Neural Edge ESP32 Classifier

Edge AI

TinyML model quantization compiled directly onto the ESP32 microcontroller dual-core chip. Classifies vibration & acoustic signals locally without requiring cloud roundtrips.

Telemetry Metrics:
Memory Footprint
84 KB RAM
Inference Time
4.2 ms
Offline Accuracy
97.3%
TinyMLESP32C++Edge AITensorFlow Lite
LiveQ4 2025

High-Frequency MQTT Mesh Broker

Distributed Protocol

Distributed micro-broker pipeline handling 50,000 sensor telemetry messages per second over WebSockets & MQTT with dynamic payload compression.

Telemetry Metrics:
Message Throughput
50k/sec
P99 Latency
1.8 ms
Packet Loss
0.00%
MQTTGoWebSocketsTimescaleDBProtobuf
ResearchQ2 2026

Adaptive Hardware Power Governor

Embedded System

Dynamic frequency scaling and power gating driver for battery-powered remote IoT sensors, extending field operational lifespan by 300%.

Telemetry Metrics:
Sleep Current
5 µA
Battery Extension
3.2x
Duty Cycle
0.1%
Low PowerFreeRTOSDeep SleepLiFePO4C++