Embedded prototypes, telemetry systems, machine-learning research, and full-stack applications—organized by what each project had to make work together.
A vehicle tracker built as one connected system—from GNSS fix to live web telemetry.
An ESP32-based 4G GNSS vehicle tracker publishing live location and telemetry over MQTTS to a custom web application.
ESP32
FreeRTOS
MQTT
GNSS
4G
FastAPI
PostgreSQL
Redis
Quick inspect +
Challenge
A useful tracker has to do more than read coordinates. It needs to keep device tasks separated, maintain a mobile data link, publish telemetry securely, expose device state, and give the backend a dependable stream to work with.
Approach
The device combines GNSS and 4G around an ESP32 running separated FreeRTOS tasks. Telemetry travels over MQTTS into a FastAPI service, with PostgreSQL and Redis supporting the application layer. An LCD, LEDs, and buzzer keep the hardware state visible while two-way messaging and OTA support remote operation.
Result
The project demonstrates an end-to-end telemetry architecture: embedded hardware, concurrent firmware, cellular networking, secure messaging, backend services, data storage, and a live web client designed as one system.
A thesis project connecting rigorous ECG model evaluation with constrained on-device inference.
A CNN-based ECG arrhythmia classifier evaluated across patients and deployed for on-device inference on ESP32.
Python
TensorFlow Lite
CNN
TinyML
ESP32
Quick inspect +
Challenge
ECG classifiers can look convincing without proving that they generalize to patients the model has not seen. The second constraint was practical: carrying the trained model from a research environment onto an ESP32-class device.
Approach
The work uses a CNN with an inter-patient evaluation framework, then converts the trained network with TensorFlow Lite for TinyML deployment. This keeps evaluation discipline and deployment constraints in the same research pipeline.
Result
The thesis links model development, inter-patient testing, TensorFlow Lite conversion, and ESP32 inference instead of stopping at an offline accuracy result.
Household power measurement, protection, and safety telemetry in one ESP32-based prototype.
Real-time household AC power monitoring with protection logic, safety sensing, and self-hosted Home Assistant integration.
ESP32
ESP-NOW
ACS712
ZMPT101B
Home Assistant
Quick inspect +
Challenge
Monitoring an AC appliance is only part of the problem. The system also needs to react to unsafe voltage or power conditions, observe temperature and gas signals, and make its state available beyond the enclosure.
Approach
Voltage and current sensing feed an ESP32 that applies over-voltage and over-power protection logic through relay control. Temperature and gas sensors extend the safety layer, while an I²C display provides local feedback and ESP-NOW carries data toward a self-hosted Home Assistant setup.
Result
The prototype brings measurement, protection, local feedback, wireless communication, and home-automation visibility together around real household AC loads.
A market platform designed around the different jobs of fishermen, wholesalers, and fisheries stakeholders.
A role-based digital fish market and fisheries advisory platform designed around price visibility and market oversight.
Next.js
TypeScript
Tailwind CSS
MySQL
JWT
Quick inspect +
Challenge
A fisheries marketplace cannot treat every participant as the same user. Trading, fair-price visibility, advisory information, and department oversight each need their own access and workflow.
Approach
The platform uses Next.js, TypeScript, MySQL, and JWT-based access to separate role-specific experiences. Market functions are paired with oversight modules and an advisory layer rather than being reduced to a generic product listing site.
Result
Machgari frames the digital fish market as a multi-stakeholder system, connecting commerce, price visibility, advisory support, and institutional oversight.
A small autonomous irrigation loop powered by solar energy and exposed through a web dashboard.
A solar-powered irrigation prototype using soil-moisture-driven pump control with web monitoring and override.
ESP32
Soil-moisture sensor
Relay
Solar
Li-ion
Quick inspect +
Challenge
The system needed to make watering decisions from soil conditions while operating from a compact solar and battery power path, without removing the operator’s ability to inspect or override it.
Approach
An ESP32 reads soil moisture and drives a pump through a relay. A solar panel, Li-ion battery, and TP4056 form the prototype power subsystem, while the web dashboard provides monitoring and manual override.
Result
The prototype combines sensing, control, renewable power, and remote supervision in a complete irrigation loop.