Software Developer & AI Researcher
Software developer currently focusing on AI research and model optimization. Experienced in building scalable full-stack applications, fine-tuning models, and contributing to open-source methodologies. Driven by complex problem-solving and technical excellence.
A custom web-based tool I developed to parse and analyze training logs from the OpenAI Parameter Golf runs. It visualizes metric trajectories over the strict 10-minute training budget, making it easier to track BPB drops and evaluate parameter efficiency. I built this to accelerate my own architectural research, and it is open for anyone to use at farcas.dev/golf.
Most of what I know is self-taught through personal projects across multiple domains. Below is a timeline of what I worked on and what I learned at each stage.
My most recent and primary focus. Actively contributing to OpenAI's Parameter Golf, drafting architectural optimizations, and working on advanced model parameter research.
My first professional role. Tasked with designing and prototyping an AI-powered voice-caller assistant for customer service.
Went beyond using AI APIs to actively developing and training models. Fine-tuned open-source LLMs for specific tasks, trained image generation models to produce custom outputs, and built RAG pipelines connecting LLMs to structured data.
Participated in penetration testing and bug bounty programs. My core approach is business logic exploitation — testing endpoints, reverse-engineering the intended server-side behavior by sending different payloads, and finding ways to use legitimate functionality in unintended ways. Also worked with OWASP methodology and network reconnaissance. Specifics are under NDA.
Built automation scripts, web scrapers, and backend services in Python. Started managing my own Linux servers — Nginx configuration, reverse proxying, load balancing across multiple upstream servers, SSL with Let's Encrypt, and Bash deployment scripts.
Learned HTML, CSS, and JavaScript, then moved into responsive layouts with Tailwind CSS and DOM manipulation.
First experience shipping a complete product. Learned C# scripting, component-based architecture, physics systems, and UI design in Unity. Also started using Git for version control.
Started by writing game logic in raw C/C++ — console-based programs, file I/O, basic data structures. Built a foundation in low-level thinking, memory management, and pointers before ever touching an engine or framework.
These are the technologies I've used across my personal projects and professional work. My core languages are highlighted below. What defines me most is contextual adaptability — I've worked across enough paradigms to pick up new tools quickly.
I've participated in a total of 5 hackathons. Here are the two most notable — along with my current active competition.
An open research challenge to train the best language model under extreme constraints: a strict 16MB limit for the entire artifact (weights + training code combined) and a 10-minute training budget on 8×H100 GPUs. The goal is to achieve the lowest bits per byte (BPB) on the FineWeb dataset. I am actively competing and drafting updates utilizing custom architectures, depth recurrence, and Quantization-Aware Training (QAT) to push parameter efficiency limits.
An AI & Business Data hackathon where participants tackled real-world challenges using machine learning, large-scale datasets, and AI tools. Featured an ML tournament for industry classification, side-quests from partners like Banca Transilvania, and access to Veridion's 115M+ company database. Grand prize: €5,000.
Beyond the highlighted events, I've participated in 3 additional hackathons across various domains — building, competing, and learning under pressure every time. Hackathons sharpen the exact skills I value most: fast execution, creative problem-solving, and delivering under tight deadlines.
ULBS — Universitatea Lucian Blaga, Sibiu
Currently 2nd yearColegiul National Ion Luca Caragiale, Moreni
Highschool GraduateBelow is my letter of recommendation from Volico Data Centers. You can view it directly or download the full CV as PDF.
Get the short PDF version of my CV — ready for printing, sharing, or submitting through application portals.
Download PDF