Program Structure

Module 1: IoT + AI Overview

• IoT architecture: device → gateway → cloud → app.
• Where AI fits: prediction, monitoring, optimization.
• IoT KPIs: uptime, latency, energy, accuracy, false alarms.
• Common issues: sensor noise, drift, missing values.

Module 2: Sensor Data Handling

• Data formats: timestamps, streams, batches.
• Cleaning: resampling, smoothing, outlier handling.
• Feature basics: rolling stats, lags, trends.
• Labeling events and building datasets.

Module 3: Time-Series Forecasting (IoT)

• Forecasting use cases: energy, temperature, demand, load.
• Train/validation splits for time-series; leakage prevention.
• Baseline models vs ML models (overview).
• Evaluation: MAE/RMSE and error analysis.

Module 4: Anomaly Detection & Alerts

• Rule-based baselines and thresholds.
• Unsupervised methods: clustering and isolation forest (intro).
• Change detection: drift and slow failures (intro).
• Alert tuning: precision vs recall; alert fatigue.

Module 5: Predictive Maintenance (Intro Workflow)

• Failure modes and signals: vibration, temperature, current, pressure.
• Classification for fault detection; early warning signals.
• Remaining Useful Life (RUL) concept and KPIs.
• Maintenance dashboards and decision rules.

Module 6: Edge AI Basics

• Why edge: latency, privacy, bandwidth, offline.
• Model choices: small classifiers and lightweight CNNs (overview).
• Compression concepts: quantization and pruning (intro).
• Edge constraints: memory, compute, power.

Module 7: Deployment & Monitoring

• Streaming pipelines: ingestion, storage, processing (overview).
• Serving models: batch vs real-time inference.
• Monitoring: drift, sensor calibration changes, performance.
• Security basics: device identity, auth, and data integrity (overview).

Module 8: End-to-End System Design

• Designing an IoT AI solution: data flow, model, alerts, UI.
• Testing: offline replay and scenario checks.
• Documentation: runbooks and failure handling.
• Cost and reliability trade-offs.

Final Project

• Pick a scenario: smart home, factory sensors, HVAC, agriculture, logistics.
• Deliverables: cleaned time-series + model + alerts + dashboard/report.
• Optional: edge deployment plan with model size targets.