About Workshop
is an advanced hands-on workshop designed to introduce learners to the deployment of lightweight artificial intelligence models on resource-constrained IoT devices. The workshop focuses on model optimization, quantization, TinyML workflows, and real-time AI inference on microcontrollers, enabling participants to understand how intelligent systems can operate efficiently at the edge without depending on cloud computing.
Aim
The aim of this workshop is to provide participants with practical knowledge of Edge AI and TinyML by teaching them how to optimize, quantize, and deploy neural network models on IoT microcontrollers for real-time, low-power, and efficient intelligent applications.
What Participants Will Learn
To introduce the fundamentals of Edge AI, TinyML, and embedded machine learning.
To explain the importance of model compression, optimization, and quantization for microcontroller-based AI deployment.
To demonstrate how neural networks can be converted into lightweight models suitable for IoT hardware.
To provide hands-on exposure to deploying AI models on resource-limited microcontrollers.
To help participants understand real-time inference, memory limitations, latency, and power-efficient AI execution.
To explore real-world applications of TinyML in smart devices, healthcare, robotics, automation, agriculture, and industrial IoT.
Structure
Day 1: Foundations of Edge AI, TinyML, and Microcontroller-Based Intelligence
- Introduction to Edge AI, TinyML, and embedded machine learning
- Difference between Cloud AI, Edge AI, and TinyML-based inference
- Role of IoT microcontrollers in real-time intelligent systems
- Architecture of TinyML workflow: data collection, model training, compression, conversion, and deployment
- Understanding hardware constraints: memory, power, latency, processing speed, and sensor limitations
- Comparing floating-point and quantized neural networks
- Applications of TinyML in smart healthcare, agriculture, industrial IoT, robotics, smart cities, and environmental monitoring
Hands-on Activity: Google Colab / Notebook
Participants will train a lightweight neural network using sample IoT/sensor-style data and evaluate its accuracy before optimization. Tools: Google Colab, Python, TensorFlow/Keras, NumPy, Pandas, MatplotlibDay 2: Quantized Neural Networks and Model Optimization for TinyML
- Why normal deep learning models cannot directly run on microcontrollers
- Introduction to model compression techniques
- Post-training quantization and quantization-aware training
- INT8 quantization and its role in reducing model size and memory usage
- Accuracy vs model size trade-off in TinyML systems
- TensorFlow Lite and TensorFlow Lite Micro workflow
- Emerging optimization methods: pruning, knowledge distillation, low-rank approximation, and hardware-adaptive neural networks
Hands-on Activity: Google Colab / Notebook
Participants will convert a trained TensorFlow/Keras model into a lightweight TensorFlow Lite model and compare performance before and after quantization. Tools: Google Colab, TensorFlow, TensorFlow Lite Converter, PythonDay 3: Deploying TinyML Models on IoT Microcontrollers and Real-Time Edge Applications
- Microcontroller deployment pipeline for TinyML applications
- Converting
.tflitemodels into microcontroller-compatible C arrays - Introduction to TensorFlow Lite for Microcontrollers
- Real-time inference workflow using sensor input
- Memory profiling, latency testing, and model performance evaluation
- TinyML use cases in biomedical signal classification, anomaly detection, predictive maintenance, smart sensing, and urban edge intelligence
- Research-based examples of real-time TinyML deployment, including biomedical and IoT sensor applications
- Future trends: secure TinyML, decentralized edge inference, on-device learning, and sustainable edge intelligence
Hands-on Activity: Google Colab / Notebook
Participants will convert a quantized.tflite model into a C-compatible model array and simulate the deployment workflow for microcontroller-based inference.
Tools: Google Colab, TensorFlow Lite, Python, NumPy, C-array conversion workflow
Important Dates
Registration Ends
4:00 PM IST
Workshop Dates
2026-08-25
05:30 PM IST
05:30 PM IST
What You Will Gain
Outcomes
- Understand the core concepts of Edge AI, TinyML, and quantized neural networks.
- Identify the challenges of deploying AI models on low-memory and low-power IoT devices.
- Optimize and quantize neural network models for embedded deployment.
- Deploy lightweight AI models on IoT microcontrollers for real-time inference.
- Evaluate model performance in terms of accuracy, latency, memory usage, and power efficiency.
- Apply TinyML techniques to build intelligent edge-based applications across research and industry domains.
Who Should Attend
Researchers working in AI, IoT, Edge Computing, Robotics, and Embedded Systems.
Academicians and faculty members exploring TinyML and Edge AI applications.
PhD scholars, postgraduate students, and research scholars in AI/ML and electronics.
Industry professionals involved in IoT, automation, smart devices, and industrial AI.
Engineers and developers interested in deploying neural networks on microcontrollers.
AI/ML learners who want practical exposure to model quantization and edge deployment.
Professionals from healthcare, agriculture, smart cities, manufacturing, and environmental monitoring.
Startups and product teams building low-power, real-time intelligent IoT solutions.