"Smart homes, smarter energy. HabitatGuard AI uses ML to predict and prevent energy waste."

Inspiration

Smart home devices are everywhere, but they operate in silos. Alexa controls lights, Ring monitors the door, and smart thermostats manage heat—but none of them talk to each other to optimize energy usage. We were inspired by the growing need for sustainable living and the gap in current smart home ecosystems that fail to "shape" user habits toward efficiency.

What It Does

HabitatGuard AI is an intelligent energy optimization layer for Amazon smart home devices. It ingests data from IoT sensors and historical usage patterns to predict energy spikes. Instead of just reacting, it proactively shapes user behavior by triggering automated routines to reduce consumption during peak hours, potentially saving users up to 15% on energy bills.

How We Built It

We built the core logic in Python, leveraging the following Amazon Developer stack:

AWS Lambda: Serverless backend to process energy data in real-time.
Amazon IoT Core: Ingests telemetry data from simulated smart home devices.
Alexa Skills Kit: Creates a voice interface for users to check their "Energy Health Score" and override automation.
Scikit-learn (Python): A lightweight ML model trained on historical usage data to predict high-consumption windows.

Challenges We Faced

The biggest challenge was simulating real-time IoT data streams for the hackathon. We had to create a robust Python script that mocked device inputs while maintaining low latency for the Alexa integration. Additionally, ensuring the ML model remained accurate with limited training data required iterative tuning of our feature selection.

Accomplishments that we’re proud of

Successfully integrated AWS IoT Core with the Alexa Skills Kit.
Built a Python-based prediction engine that achieves 85% accuracy in forecasting energy spikes.
Created a seamless voice experience where users can ask, "How am I doing with energy today?"

What We Learned

We learned how powerful the combination of AWS serverless technologies and Python AI libraries can be. We also gained deep insights into the Alexa Development Kit and how to structure skills that provide genuine utility beyond simple commands.

What’s Next for HabitatGuard AI

We plan to expand HabitatGuard to support third-party smart plugs and integrate with major utility providers to offer dynamic pricing adjustments based on real-time grid load.

Built With

alexa-skills-kit
amazon-iot-core
amazon-web-services
amazoniot
aws-lambda
machine-learning
python
scikit-learn

Submitted to

Build, Ship, Shape: Amazon Developer Hackathon

Created by

I led the end-to-end development of HabitatGuard AI, building the core Python ML prediction engine using Scikit-learn, integrating AWS Lambda and Amazon IoT Core for real-time energy data processing, and developing the Alexa Skills Kit voice interface. I designed the system architecture, implemented the IoT data simulation scripts, and trained the ML model to achieve 85% accuracy in forecasting energy spikes.

Abhishek Tiwari
Ayush Gupta