Projects

Desk Display for Environmental Monitoring

Designed and built a compact ESP8266-based desk display that provides real-time visibility of environmental data from a separate Home Assistant monitoring station. The project combines embedded development, custom UI design, iterative CAD prototyping in FreeCAD, and 3D printing to transform a breadboard prototype into a polished everyday device.

Short description:

More Details:

Embedded Display & Home Assistant Integration

Developed a compact desk display based on an ESP8266 Wemos D1 Mini, chosen for its strong price-to-performance ratio and compact form factor. The device retrieves environmental sensor data provided as entities through Home Assistant and displays temperature, humidity, CO₂ concentration, and particulate matter on a 128×128 monochrome OLED display. All relevant information is presented simultaneously on a single screen using a custom-designed layout, allowing the current room conditions to be checked at a glance.

Purpose-Built Visual CO₂ Indicator

Implemented an RGB LED as an ambient air-quality indicator based on the current CO₂ level. Instead of using a continuous color gradient, the system deliberately uses three discrete states: green, yellow, and red. This design decision makes state changes immediately noticeable in peripheral vision. A sudden transition between colors is significantly easier to recognize than a gradual color change, providing an intuitive reminder when it is time to ventilate the room.

Custom CAD Design & Iterative Prototyping

Designed the complete enclosure specifically for this project using FreeCAD, making it my first full CAD project. The design went through multiple iterations to optimize component placement, fit, and overall appearance. Rather than repeatedly printing the complete enclosure, individual components were prototyped separately to reduce material usage and iteration time.

For example, individual prototypes were created for the enclosure frame, the mounting system for the Wemos D1 Mini, and the RGB LED holder. A dedicated test piece containing multiple hole diameters was designed and printed to determine the optimal LED fit, ensuring that it remained securely mounted without being compressed or slipping.

Manufacturing Tolerances & Physical Iteration

One of the main practical challenges was compensating for differences between CAD dimensions and the actual dimensions produced by the 3D printer. Components designed with exact measurements could differ by several tenths of a millimeter after printing, requiring additional iterations and dimensional adjustments. This introduced practical experience with manufacturing tolerances and demonstrated the importance of validating digital designs against real-world physical results.

From Breadboard Prototype to Finished Product

The project originated from a functional display prototype that had already been running for several months on a breadboard. The goal was to transform this temporary setup into a compact and visually integrated device suitable for permanent use on a desk. The final enclosure hides the internal wiring and soldered connections while providing a significantly cleaner and more polished appearance than the original prototype.

Tech Stack

Embedded / IoT: ESP8266, Wemos D1 Mini, ESPHome
Integration: Home Assistant
Display: 128×128 OLED, SSD1306 over I²C
Hardware: RGB LED, soldered and wired component integration
CAD & Prototyping: FreeCAD, 3D Printing

Key Achievements

Usability: Transformed environmental data from a hidden monitoring station into an always-visible, compact desk display.
UX Design: Designed a discrete RGB state system that makes CO₂ threshold changes immediately noticeable through peripheral vision.
Rapid Prototyping: Reduced material waste and iteration time by testing individual enclosure components instead of repeatedly printing the complete design.
Engineering: Successfully compensated for real-world 3D-print manufacturing tolerances through iterative measurement and design adjustments.
Product Development: Converted a long-running breadboard prototype into a purpose-built, permanently deployed device with a custom-designed enclosure.

CI/CD Pipeline and Infrastructure as Code

Automated a full CI/CD pipeline using GitHub Actions and Watchtower for containerized deployments. I modularized the infrastructure by migrating to a version-controlled 'Service Include' architecture, significantly reducing manual overhead and improving system maintainability.

Short description:

More Details:

CI/CD Pipeline & Container Orchestration

Implemented a streamlined pipeline using GitHub Actions. Upon merging to the main branch, the system triggers an automated build process, containerizes the application, and publishes the image to the GitHub Container Registry (GHCR).

Security-First Development (DevSecOps):

Integrated Trivy into the CI workflow to perform automated vulnerability scans on container images and dependencies. The pipeline employs a fail-fast methodology: if a high-risk vulnerability or CVE is detected, the deployment is automatically aborted, ensuring that only hardened, secure code reaches the production environment.

Automated Deployment & Zero-Downtime Strategy

Utilized Watchtower for automated image polling and lifecycle management on the remote VPS. Combined with Traefik as a dynamic Reverse Proxy, the architecture allows for seamless updates and the deployment of new microservices via subdomains without interrupting existing traffic.

Infrastructure as Code (IaC) & Versioning

The entire server infrastructure is managed as code within a dedicated repository. By using a centralized Docker Compose structure with modular includes, the environment remains highly maintainable and version-controlled, allowing for instant rollbacks and reproducible environments.

Real-time Monitoring & Incident Response

Configured automated alerting to provide instant mobile notifications for pipeline failures, ensuring high availability and rapid MTTR (Mean Time To Recovery).

Tech Stack

CI/CD / DevOps: GitHub Actions, GHCR, Watchtower
Security: Trivy (Vulnerability Scanning)
Infrastructure: Docker, Docker Compose, Traefik (Reverse Proxy)

Key Achievements

Efficiency: Reduced deployment time from minutes of manual CLI work to zero-touch automation.
Security: Blocked 100% of known vulnerable dependencies from reaching production through automated gating.
Scalability: Modular configuration allows adding new services with a single line of code in the global compose file.

Automotive IoT: Reverse-Engineering & Telemetry Integration

Extracted real-time vehicle telemetry by reverse-engineering a proprietary CAN-Bus network through a 100% reversible hardware harness. Developed a secure WireGuard VPN and MQTT pipeline to stream live diagnostics into a Home Assistant smart dashboard, leveraging LLMs for automated payload analysis.

Short description:

More Details:

Physical Layer & Hardware Integration

Designed and implemented a 100% non-destructive, reversible hardware installation using a custom Quadlock passthrough adapter behind the vehicle's head unit. This allowed safe, isolated access to the 12V, GND, CAN High, and CAN Low communication lines without altering factory wiring. The signals were routed to a secondary OBD-II port acting as a physical-to-digital bridge via a WiCAN Pro hardware transceiver.

Smart Reverse Engineering & Differential Analysis

Replaced traditional, time-consuming manual byte monitoring with an advanced AI-assisted engineering workflow. Baseline CAN frame dumps were captured under varying operational states and processed using a Large Language Model (LLM) for pattern recognition. This successfully isolated proprietary CAN IDs, endianness (Little Endian), byte configurations, and scaling factors. Hardware-level acceptance masks and filters were then deployed to block bus noise at the edge.

Secure WAN Telemetry & Network Automation

Established a robust edge-to-cloud wide-area network path to ensure continuous data transmission outside local Wi-Fi range. An automated script triggers a smartphone hotspot instantly upon Apple CarPlay connection. The WiCAN Pro automatically connects to this link, routing lightweight MQTT telemetry payloads through an encrypted WireGuard VPN tunnel on a public VPS directly to the local home automation instance.

Full-Stack IoT Dashboard & Parsing

Configured custom YAML parsing at the ingestion layer to convert raw, filtered CAN-Bus payloads into accurate entity states and functional variables. The resulting telemetry metrics (such as fuel level and odometer readings) are displayed in real-time on a localized Home Assistant dashboard, enabling seamless system automation and historical data tracking.

Tech Stack

Hardware & Protocols: CAN-Bus, OBD-II, WiCAN Pro, Custom Quadlock Passthrough
Networking & Security: WireGuard VPN, VPS, MQTT, Edge-Filtering / Masking
Automation & Analysis: Home Assistant, LLM Pattern Recognition, Apple CarPlay / iOS Shortcuts

Key Achievements

Innovation: Accelerated proprietary protocol decoding by combining differential data harvesting with LLM pattern recognition.
Security & Protection: Ensured complete vehicle asset protection with a 100% reversible physical installation and strict WireGuard network encapsulation.
Full-Stack Fluency: Successfully bridged the gap between low-level automotive physical layers, secure cloud networking, and high-level home automation ecosystems.

ERM - Editor

I have developed an editor for entity relation ship diagrams. With d3.js i made sure that the elements that make up the diagram are automatically arranged appropriately, making it extremely easy to work on a diagram. You can try it out here.

Control lamp over WIFI

In this project i created a small library to control a lamp wirelessly using Python code over a WiFi network. With simple code commands, users can turn the lamp on/off, adjust brightness levels, change the color and schedule automatic operations. This project offers convenience, flexibility, and automation possibilities for lamp control within a network environment.

Spotify Stats PWA

In this project, I developed a Progressive Web Application (PWA) that integrates with Spotify by using its API. By connecting their Spotify account, users can track their listening habits, and the app provides detailed statistics on listening patterns, including the times when they typically listen to their favorite songs.

Fire detection AI

In this project, I developed an AI system to detect fires in images using advanced machine learning techniques. The system utilizes a convolutional neural network trained on a diverse dataset to accurately identify fire patterns, offering a promising solution for enhancing safety and efficiency in fire detection across various domains.

AI Terminal

This project focuses on developing a console application that leverages OpenAI's GPT API to enhance the usability of the Windows Command Prompt (CMD) by integrating natural language processing capabilities. The application enables users to interact with CMD using natural language commands, eliminating the need to memorize specific command syntax.

Progressive Web Application

I developed a Progressive Web Application (PWA) that provides real-time statistics for the popular game Valorant. Users can access comprehensive statistics, including player rankings, match histories or live data. They can also perform certain actions from within the app.

Wikipedia Redesign

This project introduces a browser extension designed to modernize the appearance of Wikipedia. By implementing aesthetic enhancements, such as updated fonts and layout adjustments, the extension aims to provide users with a more visually appealing and contemporary browsing experience on the platform while staying true to its original design. This project offers a simple yet effective solution for those seeking to refresh the interface of Wikipedia, enhancing usability and engagement for readers and researchers alike.

Mailclient

This project involved crafting an email client using HTML, CSS, and JavaScript for the frontend interface, while employing Python for the backend functionality. The client enables users to interact with their email accounts via SMTP and IMAP protocols.

File explorer

This project introduces a user-friendly solution for accessing and managing files across devices within a network. With the intuitive interface users can seamlessly navigate and control their files from any other device in the same network. This project aims to enhance productivity and streamline file management within network environments.

Spotify Redesign

This design project focuses on reimagining the visual experience of Spotify for users. The aim is to enhance user engagement and satisfaction by creating a more intuitive, aesthetically pleasing, and personalized interface.