vinod sharma .in Solution Architect, Author & Educator
Courses, books, roadmaps, and tutorials to help developers build real-world skills.
© 2026 Vinod Sharma. All rights reserved.
Edge Computing Engineer Build intelligent systems that process data where it's generated
10 milestones in this roadmap
Step 1 beginner 6-8 weeks
Embedded Systems Basics Build hands-on embedded programming skills that enable you to work directly with the hardware powering edge computing deployments.
1 Microcontroller architecture: CPU cores, memory hierarchy (Flash, SRAM), and peripheral buses 2 GPIO programming: digital I/O, pull-up/pull-down resistors, debouncing, and pin multiplexing 3 Interrupt handling: ISR design, priority levels, nested interrupts, and DMA transfers 4 Memory-mapped I/O: register access, bitfield manipulation, and hardware abstraction layers 5 RTOS concepts: tasks, scheduling algorithms, semaphores, mutexes, and inter-task communication 6 ARM architecture: Cortex-M series, instruction set, power modes, and debugging with SWD/JTAG
Arduino IDE STM32CubeIDE FreeRTOS PlatformIO ARM Keil MDK
Step 1 beginner 6-8 weeks
Embedded Systems Basics Build hands-on embedded programming skills that enable you to work directly with the hardware powering edge computing deployments.
1 Microcontroller architecture: CPU cores, memory hierarchy (Flash, SRAM), and peripheral buses 2 GPIO programming: digital I/O, pull-up/pull-down resistors, debouncing, and pin multiplexing 3 Interrupt handling: ISR design, priority levels, nested interrupts, and DMA transfers 4 Memory-mapped I/O: register access, bitfield manipulation, and hardware abstraction layers 5
Step 2 beginner 4-6 weeks
IoT Protocols & Communication Understand the communication landscape for edge devices and learn to select the right protocol based on range, bandwidth, power, and latency requirements.
1 MQTT: publish-subscribe model, QoS levels (0/1/2), retained messages, last will, and broker clustering 2 CoAP: RESTful constrained protocol, observe pattern, block-wise transfer, and DTLS security 3 BLE: GATT profiles, advertising, connection intervals, bonding, and mesh networking 4
Step 3 intermediate 4-6 weeks
Edge Hardware Platforms Develop practical experience with the hardware ecosystem for edge computing, learning to match platforms to performance, power, and cost constraints.
1 Raspberry Pi: Linux-based computing, camera interface, GPIO expansion, and cluster configurations 2 NVIDIA Jetson: GPU-accelerated edge AI, CUDA cores, JetPack SDK, and DeepStream pipeline 3 Intel NUC: x86 edge server, vPro management, and enterprise-grade edge deployments 4 Google Coral: Edge TPU architecture, model compilation, and USB/PCIe accelerator modules
Step 4 intermediate 4-6 weeks
Containerization at the Edge Apply cloud-native container patterns to edge computing with adaptations for limited resources, intermittent connectivity, and fleet management.
1 Lightweight container runtimes: containerd, CRI-O, and Podman for resource-constrained environments 2 K3s: lightweight Kubernetes distribution, single-binary deployment, and edge-optimised networking 3 MicroK8s: snap-based Kubernetes, add-on ecosystem, and high-availability clustering 4 Container registries: local registries, image caching, and bandwidth-optimised distribution
Step 5 intermediate 6-8 weeks
Edge AI & TinyML Bridge the gap between cloud AI and edge deployment by learning to optimise models for devices with severe compute, memory, and power constraints.
1 Model compression: knowledge distillation, weight sharing, and architecture search for edge models 2 Quantisation: post-training quantisation (INT8/INT4), quantisation-aware training, and mixed precision 3 Pruning: structured vs unstructured pruning, magnitude-based pruning, and iterative pruning schedules 4
Step 6 intermediate 4-6 weeks
Data Processing at Edge Design data architectures that process and filter data at the edge, reducing bandwidth costs and latency while maintaining cloud synchronisation.
1 Stream processing at edge: Apache Flink Stateful Functions, Apache NiFi MiNiFi, and edge event processing 2 Time-series databases: InfluxDB, TimescaleDB, and QuestDB for sensor data storage 3 Local analytics: aggregation, anomaly detection, and statistical analysis on constrained devices 4 Data filtering: edge data reduction, sampling strategies, and change-data-capture at the edge
Step 7 advanced 4-6 weeks
Edge Security Protect edge deployments against the unique security threats of distributed, physically accessible, and often unattended computing devices.
1 Device authentication: X.509 certificates, device identity, and mutual TLS for device-to-cloud 2 Secure boot: chain of trust, bootloader verification, and signed kernel images 3 Firmware signing: code signing certificates, signature verification, and anti-rollback protection 4 OTA security: encrypted update packages, differential updates, and secure update channels
Step 8 advanced 6-8 weeks
Edge Orchestration Master the platforms and practices needed to deploy, manage, and troubleshoot thousands of edge devices across geographically distributed locations.
1 Fleet management: device grouping, configuration management, and staged rollout strategies 2 Device provisioning: zero-touch provisioning, device registration, and identity management 3 Remote debugging: remote shell access, log collection, diagnostic commands, and crash analysis 4 AWS Greengrass: local Lambda functions, machine learning inference, and stream management
Step 9 advanced 6-8 weeks
Digital Twins Create virtual representations of physical assets that enable simulation, monitoring, and prediction without risking the physical systems.
1 Virtual replica design: entity modeling, property mapping, and state synchronisation architecture 2 Simulation models: physics-based simulation, discrete event simulation, and agent-based modeling 3 Real-time sync: telemetry ingestion, bi-directional updates, and latency requirements 4 Predictive maintenance: anomaly detection, remaining useful life estimation, and maintenance scheduling
Step 10 advanced 4-6 weeks
Industry Applications Understand how edge computing transforms industries and learn to design solutions tailored to specific domain requirements and regulatory environments.
1 Smart manufacturing: predictive quality, real-time defect detection, and OPC-UA integration 2 Autonomous vehicles: sensor fusion, V2X communication, and real-time decision making at the edge 3 Retail analytics: computer vision for footfall counting, shelf monitoring, and checkout-free stores 4 Healthcare monitoring: wearable sensor processing, FDA compliance, and HIPAA-compliant edge processing Ready to start this journey? Browse our courses and books to begin your learning path.
RTOS concepts: tasks, scheduling algorithms, semaphores, mutexes, and inter-task communication
6 ARM architecture: Cortex-M series, instruction set, power modes, and debugging with SWD/JTAG
Arduino IDE STM32CubeIDE FreeRTOS PlatformIO ARM Keil MDK
LoRaWAN: chirp spread spectrum, adaptive data rate, device classes (A/B/C), and network architecture
5 Zigbee: mesh topology, coordinator/router/end-device roles, and Zigbee 3.0 clusters
6 Protocol selection criteria: power budget, range, data rate, latency, and network topology trade-offs
Mosquitto (MQTT broker) nRF Connect (BLE) The Things Network (LoRaWAN) Wireshark IoT dissectors
Step 2 beginner 4-6 weeks
IoT Protocols & Communication Understand the communication landscape for edge devices and learn to select the right protocol based on range, bandwidth, power, and latency requirements.
1 MQTT: publish-subscribe model, QoS levels (0/1/2), retained messages, last will, and broker clustering 2 CoAP: RESTful constrained protocol, observe pattern, block-wise transfer, and DTLS security 3 BLE: GATT profiles, advertising, connection intervals, bonding, and mesh networking 4 LoRaWAN: chirp spread spectrum, adaptive data rate, device classes (A/B/C), and network architecture 5 Zigbee: mesh topology, coordinator/router/end-device roles, and Zigbee 3.0 clusters 6 Protocol selection criteria: power budget, range, data rate, latency, and network topology trade-offs
Mosquitto (MQTT broker) nRF Connect (BLE) The Things Network (LoRaWAN) Wireshark IoT dissectors
5 ESP32: dual-core WiFi/BLE SoC, deep sleep modes, ULP coprocessor, and ESP-IDF framework
6 Hardware selection: compute requirements, power envelope, connectivity, cost, and thermal constraints
Raspberry Pi OS NVIDIA JetPack SDK ESP-IDF / Arduino Google Coral Edge TPU Compiler
Step 3 intermediate 4-6 weeks
Edge Hardware Platforms Develop practical experience with the hardware ecosystem for edge computing, learning to match platforms to performance, power, and cost constraints.
1 Raspberry Pi: Linux-based computing, camera interface, GPIO expansion, and cluster configurations 2 NVIDIA Jetson: GPU-accelerated edge AI, CUDA cores, JetPack SDK, and DeepStream pipeline 3 Intel NUC: x86 edge server, vPro management, and enterprise-grade edge deployments 4 Google Coral: Edge TPU architecture, model compilation, and USB/PCIe accelerator modules 5 ESP32: dual-core WiFi/BLE SoC, deep sleep modes, ULP coprocessor, and ESP-IDF framework 6 Hardware selection: compute requirements, power envelope, connectivity, cost, and thermal constraints
Raspberry Pi OS NVIDIA JetPack SDK ESP-IDF / Arduino Google Coral Edge TPU Compiler
5 OTA updates: container image updates, rollback strategies, and A/B partition schemes
6 Resource constraints: memory limits, CPU quotas, and optimising container images for ARM architectures
K3s / K3OS MicroK8s Docker / containerd Balena (fleet management) Portainer
Step 4 intermediate 4-6 weeks
Containerization at the Edge Apply cloud-native container patterns to edge computing with adaptations for limited resources, intermittent connectivity, and fleet management.
1 Lightweight container runtimes: containerd, CRI-O, and Podman for resource-constrained environments 2 K3s: lightweight Kubernetes distribution, single-binary deployment, and edge-optimised networking 3 MicroK8s: snap-based Kubernetes, add-on ecosystem, and high-availability clustering 4 Container registries: local registries, image caching, and bandwidth-optimised distribution 5 OTA updates: container image updates, rollback strategies, and A/B partition schemes 6 Resource constraints: memory limits, CPU quotas, and optimising container images for ARM architectures
K3s / K3OS MicroK8s Docker / containerd Balena (fleet management) Portainer
TensorFlow Lite: model conversion, delegate APIs (GPU, NNAPI, Edge TPU), and on-device training
5 ONNX Runtime: model interoperability, execution providers, and cross-platform inference
6 Edge inference optimisation: batch processing, model pipelining, and hardware-specific acceleration
TensorFlow Lite ONNX Runtime NVIDIA TensorRT Edge Impulse Apache TVM
Step 5 intermediate 6-8 weeks
Edge AI & TinyML Bridge the gap between cloud AI and edge deployment by learning to optimise models for devices with severe compute, memory, and power constraints.
1 Model compression: knowledge distillation, weight sharing, and architecture search for edge models 2 Quantisation: post-training quantisation (INT8/INT4), quantisation-aware training, and mixed precision 3 Pruning: structured vs unstructured pruning, magnitude-based pruning, and iterative pruning schedules 4 TensorFlow Lite: model conversion, delegate APIs (GPU, NNAPI, Edge TPU), and on-device training 5 ONNX Runtime: model interoperability, execution providers, and cross-platform inference 6 Edge inference optimisation: batch processing, model pipelining, and hardware-specific acceleration
TensorFlow Lite ONNX Runtime NVIDIA TensorRT Edge Impulse Apache TVM
5 Edge-to-cloud sync: store-and-forward patterns, conflict resolution, and eventual consistency
6 Data pipeline design: buffering, backpressure, and graceful degradation during connectivity loss
Apache NiFi MiNiFi InfluxDB / TimescaleDB Apache Kafka (edge) AWS IoT Greengrass streams
Step 6 intermediate 4-6 weeks
Data Processing at Edge Design data architectures that process and filter data at the edge, reducing bandwidth costs and latency while maintaining cloud synchronisation.
1 Stream processing at edge: Apache Flink Stateful Functions, Apache NiFi MiNiFi, and edge event processing 2 Time-series databases: InfluxDB, TimescaleDB, and QuestDB for sensor data storage 3 Local analytics: aggregation, anomaly detection, and statistical analysis on constrained devices 4 Data filtering: edge data reduction, sampling strategies, and change-data-capture at the edge 5 Edge-to-cloud sync: store-and-forward patterns, conflict resolution, and eventual consistency 6 Data pipeline design: buffering, backpressure, and graceful degradation during connectivity loss
Apache NiFi MiNiFi InfluxDB / TimescaleDB Apache Kafka (edge) AWS IoT Greengrass streams
5 Physical security: tamper detection, secure enclosures, and data-at-rest encryption
6 Hardware security modules: TPM 2.0, secure elements, and ARM TrustZone for key storage
TPM 2.0 tools OpenSSL / mbed TLS UEFI Secure Boot ARM TrustZone HashiCorp Vault (edge agent)
Step 7 advanced 4-6 weeks
Edge Security Protect edge deployments against the unique security threats of distributed, physically accessible, and often unattended computing devices.
1 Device authentication: X.509 certificates, device identity, and mutual TLS for device-to-cloud 2 Secure boot: chain of trust, bootloader verification, and signed kernel images 3 Firmware signing: code signing certificates, signature verification, and anti-rollback protection 4 OTA security: encrypted update packages, differential updates, and secure update channels 5 Physical security: tamper detection, secure enclosures, and data-at-rest encryption 6 Hardware security modules: TPM 2.0, secure elements, and ARM TrustZone for key storage
TPM 2.0 tools OpenSSL / mbed TLS UEFI Secure Boot ARM TrustZone HashiCorp Vault (edge agent)
5 Azure IoT Edge: module deployment, message routing, and offline operation capabilities
6 KubeEdge: cloud-edge coordination, EdgeMesh networking, and device mapper framework
AWS IoT Greengrass Azure IoT Edge KubeEdge Balena Cloud EdgeX Foundry
Step 8 advanced 6-8 weeks
Edge Orchestration Master the platforms and practices needed to deploy, manage, and troubleshoot thousands of edge devices across geographically distributed locations.
1 Fleet management: device grouping, configuration management, and staged rollout strategies 2 Device provisioning: zero-touch provisioning, device registration, and identity management 3 Remote debugging: remote shell access, log collection, diagnostic commands, and crash analysis 4 AWS Greengrass: local Lambda functions, machine learning inference, and stream management 5 Azure IoT Edge: module deployment, message routing, and offline operation capabilities 6 KubeEdge: cloud-edge coordination, EdgeMesh networking, and device mapper framework
AWS IoT Greengrass Azure IoT Edge KubeEdge Balena Cloud EdgeX Foundry
5 NVIDIA Omniverse: USD-based simulation, physics engine integration, and photorealistic rendering
6 Azure Digital Twins: DTDL modeling language, twin graphs, event routing, and integration with IoT Hub
Azure Digital Twins NVIDIA Omniverse AWS IoT TwinMaker Bentley iTwin Unity Reflect
Step 9 advanced 6-8 weeks
Digital Twins Create virtual representations of physical assets that enable simulation, monitoring, and prediction without risking the physical systems.
1 Virtual replica design: entity modeling, property mapping, and state synchronisation architecture 2 Simulation models: physics-based simulation, discrete event simulation, and agent-based modeling 3 Real-time sync: telemetry ingestion, bi-directional updates, and latency requirements 4 Predictive maintenance: anomaly detection, remaining useful life estimation, and maintenance scheduling 5 NVIDIA Omniverse: USD-based simulation, physics engine integration, and photorealistic rendering 6 Azure Digital Twins: DTDL modeling language, twin graphs, event routing, and integration with IoT Hub
Azure Digital Twins NVIDIA Omniverse AWS IoT TwinMaker Bentley iTwin Unity Reflect
5 Precision agriculture: drone-based imaging, soil sensor networks, and automated irrigation control
6 Cross-industry patterns: reference architectures, compliance frameworks, and ROI measurement
AWS Panorama (retail/industrial) NVIDIA Metropolis ROS 2 (robotics/AV) OpenCV (edge vision)
Step 10 advanced 4-6 weeks
Industry Applications Understand how edge computing transforms industries and learn to design solutions tailored to specific domain requirements and regulatory environments.
1 Smart manufacturing: predictive quality, real-time defect detection, and OPC-UA integration 2 Autonomous vehicles: sensor fusion, V2X communication, and real-time decision making at the edge 3 Retail analytics: computer vision for footfall counting, shelf monitoring, and checkout-free stores 4 Healthcare monitoring: wearable sensor processing, FDA compliance, and HIPAA-compliant edge processing 5 Precision agriculture: drone-based imaging, soil sensor networks, and automated irrigation control 6 Cross-industry patterns: reference architectures, compliance frameworks, and ROI measurement
AWS Panorama (retail/industrial) NVIDIA Metropolis ROS 2 (robotics/AV) OpenCV (edge vision)