Razvan-Constantin Tudorache

Autopilot direct-to transitions, X-Crafts ERJ / E-Jets

On volunteer for the X-Crafts team. Direct-to waypoint transitions in the simulator are not smooth: the aircraft overshoots or snaps instead of arcing. I am tracing the autopilot path in C++ through the X-Plane SDK to find where the heading intercept logic diverges from the real aircraft’s behavior, then feeding the fix back upstream.

WinWing CDU scripting for MobiFlight

Keeping the Rotate MD-11 CDU script dynamic as MobiFlight evolves. Working with the maintainers on edge cases other pilots hit with the same hardware.

Rotate MD-11 CDU driver for WinWing hardware

The Rotate MD-11 has a real, physical CDU you can own. I built a Python driver that reads live flight data from the X-Plane 12 API (position, route, performance, fuel), parses it, and draws it on a WinWing CDU so pilots can fly the aircraft with real hardware instead of a mouse.

MobiFlight had no script for the Rotate MD-11 CDU. The stock approach was hard-coded for each aircraft, so I wrote a dynamic implementation that works with every route loaded into the aircraft.

What was hard: the data parsing. X-Plane returns different answer formats depending on aircraft state, and the CDU layout expects data in strict positions. My first attempts worked for my own flight but broke for other pilots. It took several review rounds with the MobiFlight maintainers to make the script dynamic, with variable page definitions, runtime field mapping, and no hard-coded waypoints.

What worked: treating the data layer and the display layer as separate concerns, then letting the community drive the edge cases during review. The script is now part of the MobiFlight repository and works for other MD-11 pilots.

This site: Django and Coolify on my own server

I built this portfolio with Django, Tailwind CSS, and a server-side GitHub contribution display. I then prepared a Fedora Server mini PC, installed Coolify, configured DNS and router port forwarding, and secured administration with key-only SSH, firewalld, and Fail2ban.

The deployment uses a multi-stage Dockerfile that compiles the frontend, installs Python dependencies, runs migrations, collects static files, and starts Gunicorn. Coolify pulls from GitHub, manages TLS and environment variables, mounts persistent storage, and checks each new container before sending traffic to it.