Role: Goat Herder
Passions: Game Development, Machine Learning, Geeky Stuff
Description: A Software Engineer with TONS of ideas and a sweet spot for Video Games ᕕ( ᐛ )ᕗ
Things I’m proud about
👾 Publishing my first game
In 2024 I actually finished a game which was then later used to promote the HEITS Apuseni Ski Cup. You ca actually play the game in the browser, by accessing this link.
🎥 My first YouTube video
In 2023 I posted my first YouTube video in which I’m presenting how I trained a ML model to play Mario Kart. It’s not much, but it’s honest work.
⌨️ Re-designed a 5000+ line codebase
In 2022 me and a colleague had the task to re-write a giant Java monolith application into Python microservices. My colleague knew Java, I knew Python and we made it 4 times faster, made the app easier to scale and upgrade, and reduced monthly costs from 2000. Did I mention we did it in just 5 months?
📬 My first blogpost
In 2022 I made my first public technical blogpost. It may not seem like much, but it gave me the confidence boost to create this website. It is also the most clicked blogpost on the HEITS website (as of 2024).
🏃♀️ Speedrun - Publish a patent in the first 5 months
In 2021 after changing jobs and being employed for only a few months, I saw the need for more synthetic data, this time generated by a GAN (that’s like the grandfather of Deep Fakes). Seeing how fast Generative Networks progressed, I rushed to patent the idea before leaving the company.
📃 Publishing my first ever patent
In 2020 while working at Porsche, we came up with a cool idea to easily create synthetic data from real-life recordings to help us better train Machine Learning models. So we said why not, and patented the idea.
🏎️ Driving a real car using my own ML model
Around 2018 while working at Porsche Engineering Cluj, I’ve managed to deploy a Machine Learning based Adaptive Cruise Controller on a real Porsche car. The controller read the signals from the Flexray and CAN busses and overwrote the original ACC outputs with the outputs of a Neural Network. The network was trained from scratch, first through Supervised Learning and Reinforcement Learning.
🐍 Using a spinning blade as a game display
In 2017 me and a friend made the jankiest Persistence of Vision display for the Digilent Design Contest. It was powered by an FPGA and you could play Snake on it. The POV was made from a CPU fan, a plastic bowl and 3D printed propeller (that we balanced using fishing weights). We got 4th place, but we felt like absolute legends.