Notes from the studio.

Unify Z-Wave, Zigbee, WiFi, and Modbus devices in a single Home Assistant instance with a real-time Energy Sankey Diagram visualizing power flow between solar, grid, home loads, and EV charger.

Build an acoustic anomaly detection system using YAMNet TensorFlow Lite on Raspberry Pi 5 to detect glass breaking, screaming, and alarm sounds, triggering Home Assistant alerts and LINE notifications.

Install SCD41 NDIR CO2 sensor to control ERV with Demand-Controlled Ventilation, maintaining CO2 below 1,000 ppm during sleep for measurable sleep quality improvement while reducing energy consumption.

Use Apple RoomPlan API to generate a 3D room model on iPhone, then overlay AR visualizations of structural walls, MEP routing, and renovation layouts to validate design decisions before demolition.

Design a Demand Response automation aligned with MEA TOU electricity rates, using pre-cooling before peak hours and appliance load shifting to reduce electricity bills by 20–35% without affecting comfort.

Design a wearable fall detection system for elderly using MPU6050 IMU on ESP32 with SVM algorithm and context-aware suppression to distinguish real falls from normal sit-to-stand activity.

Design a Node-RED automation engine that learns occupancy patterns, detects conflicts between overlapping automation scenes, and resolves them automatically with full audit logging.

Deploy InsightFace on NVIDIA Jetson Nano for face-recognition gate and door control with anti-spoofing liveness detection, processing all biometric data locally without cloud transmission.

Build a sensor fusion system combining SDS011 PM2.5, BME680 TVOC, and SCD40 CO2 readings through a Kalman Filter to compute an accurate Indoor Air Quality Index with automated LINE alerts.

Use iPhone LiDAR to scan exterior spaces and gardens, generating PointCloud meshes imported into Twinmotion for landscape design, sun path simulation, and plant placement planning.

Build a Presence Simulation Engine on Home Assistant using randomized lighting, audio, and Markov Chain scheduling to convincingly mimic occupancy and deter burglars while away.

Design ventilation systems to maintain positive indoor pressure, preventing PM2.5 and outdoor pollutants from infiltrating through building envelope gaps during high AQI events.

Design a comprehensive elderly medication management system from biometric-verified dispensing through vital sign side effect monitoring with automated physician reporting via LINE.

Non-Intrusive Load Monitoring (NILM) analyzes the main meter waveform to identify per-appliance energy consumption without installing individual sensors on every device.

Detect pre-intrusion behavioral anomalies through zone loitering duration analysis, unusual vehicle parking patterns, and exterior lighting anomalies using Frigate NVR integrated with Home Assistant.

Convert Revit models into standalone VR experiences with Enscape and Twinmotion, enabling clients to review and approve designs before construction begins.

Design a smart flood and fire early warning system using Zigbee water leak sensors under sinks and washing machines, smoke detectors (Nest Protect vs Zigbee), CO detectors, and automated fire response that shuts HVAC and unlocks exit doors — with immediate LINE alerts.

Use pressure mat arrays or optical flow from cameras to analyse elderly gait — detecting gait speed below 0.8 m/s and shortened step length as early fall risk biomarkers, with a digital TUG test via Home Assistant alerting families before accidents occur.

Design a unified smart home dashboard architecture using Mosquitto MQTT broker, Node-RED for data flow, InfluxDB time-series database, and Grafana for visualisation — consolidating security, air quality, and elderly monitoring data into a single dashboard.

Design a cellular backup communicator for smart security using 4G LTE (SIM7600) and GSM (SIM800L) modules to deliver SMS/call alerts when fibre internet or Wi-Fi fails — ensuring 99.9% uptime even if an intruder cuts the line.

Use an LSTM neural network to forecast Bangkok AQI 1–4 hours ahead using IQAir API data, meteorological inputs, and seasonal patterns — proactively controlling air purifiers and ERV before PM2.5 reaches hazardous levels, reducing average exposure by 40–60%.

Use AI to generate interior design concepts with RoomGPT/Decor8.ai from real room photos — comparing AR interior apps (IKEA Place, Houzz, Planner 5D) and Apple Vision Pro for spatial interior review before purchasing furniture.

Build a fully local voice assistant on Raspberry Pi 5 using Whisper.cpp (STT), Ollama + LLaMA 3.2 (local LLM), and Piper TTS — comparing privacy, latency, and monthly cost against Google Assistant and Alexa for the Thai market.

Build a passive elderly health monitoring system requiring no wearables — using 60GHz mmWave radar for vital signs, toilet visit frequency tracking, Withings Body+ BIA smart scale, IoT pill dispenser, and ML anomaly detection with LINE family alerts.