NEXURA LABS
← All work

EDGE + CLOUD + AI ACADEMIC IOT PROJECT

Smart Feeder.

From a feeding command to a cloud-connected system.

An AWS-connected IoT and AI pet feeding analytics platform, evolved from an ESP32 feeder controlled through Telegram and MQTT.

Explore architecture ↓Prototype demo ↗GitHub ↗
03 / SMART FEEDER
Original Smart Feeder physical project prototype from its public repository
EDGE → CLOUD → INSIGHTA system starts
in the real world.
Original physical prototype · Project repository media
UniSZA MPI
Top 3 / Bronze

01

A physical problem, a layered solution.

The original system connects an ultrasonic food-level sensor and feeder motor to an ESP32. A Telegram bot accepts feed and food-level commands, with MQTT connecting application commands to the device.

This earlier implementation is what the public repository currently exposes. The later iteration expands the idea into camera-assisted pet identification and cloud data processing.

02

The cloud architecture evolved.

The expanded version combines ESP32-CAM, Python/FastAPI and MobileNetV2 for pet recognition. Amazon S3 stores registered pet images; DynamoDB holds feeding logs, daily statistics and pet information.

AWS Lambda supports processing and aggregation workflows. Amazon Bedrock provides AI-assisted analysis based on feeding data. These cloud-side components are described as the evolved architecture, not as fully published source code.

03

What the prototype taught me.

Working across edge hardware, messaging, application logic and cloud data made the boundaries between components concrete. A feeding command and a generated insight need different processing paths.

Top 3 / Bronze — UniSZA MPI. The prototype and expanded architecture are presented with their respective scope intact.

COMPONENTS & CONNECTIONS

System architecture.

SYSTEM OVERVIEWConceptual flow →

Control at the edge

  1. client
    TelegramPet-owner commands
  2. iot
    MQTT / ESP32Sensor + feeder motor
  3. iot
    ESP32-CAMExpanded camera iteration

Identify & persist

  1. ai
    FastAPIMobileNetV2 recognition
  2. cloud
    Amazon S3Registered pet images
  3. data
    DynamoDBLogs, statistics, pet records

Process & interpret

  1. cloud
    AWS LambdaProcessing + aggregation
  2. ai
    Amazon BedrockAI-assisted feeding analysis
  3. client
    Pet ownerGenerated insights
Smart Feeder — This is the expanded AWS/AI architecture. The public repository contains the earlier IoT/Telegram implementation and media; not every later cloud component is published.

TECHNOLOGY & INTEGRATIONS

Each part has a purpose.

  • ESP32
  • Python
  • FastAPI
  • AWS Lambda
  • DynamoDB
  • Bedrock
Amazon S3
Registered pet image dataset
Amazon DynamoDB
Feeding events, daily stats and pet records
AWS Lambda
Serverless processing and aggregation workflows
Amazon Bedrock
AI-assisted feeding and health analysis, not a veterinary diagnosis

Project scope and implementation references: view the repository ↗. Recognition, professional context and historical metrics are supplied by the project owner.

Continue the conversation.

Interested in the system, the decisions or what comes next?

Get in touch →
NEXT PROJECT / 04KnotYet