● AI/ML Engineer & PhD Researcher in AI
Sebastian
Buzdugan
“PhD-grade AI insight, builder-friendly delivery.”
I break down what new AI papers and tools mean for engineers who ship — paper breakdowns, hands-on tool guides (Claude Code, MCP, RAG), and engineering-grade commentary. Also building @getfrai, open-source tooling for responsible AI.
Research
PhD in AI · URV
Building
@getfrai · frai.cc
Publishing
@sebuzdugan
Open source
FRAI toolkit
What I do
Paper breakdowns
What the paper said, what the benchmark actually shows, and what it changes for production systems.
Tool guides
Hands-on, end-to-end builds with Claude Code, MCP servers, and RAG stacks — the way you'd actually ship them.
AI engineering
Agents, retrieval, and infrastructure — opinionated, builder-first commentary on what holds up in production.
Responsible AI research
PhD work on deep learning for medical imaging, plus FRAI: tooling for the EU AI Act, NIST AI RMF and ISO/IEC 42001.
Selected work
Open-source toolkit for responsible AI: CLI + SDK to scan code, collect evidence, and generate model cards, risk files, evals, and RAG indexes.
FRAI Benchmark: Future Responsible AI Evaluation A consensus-based safety and compliance benchmark for SOTA LLMs (DeepSeek, Grok, GPT-5). Uses a "Panel of Experts" methodology where multiple frontier models judge response quality to ensure non-biased, high-fidelity safety scores
A practical prompt library for GPT-5.5, Claude Opus 4.7, Gemini, and modern AI workflows
Responsible-AI copilot grounded in EU AI Act, NIST AI RMF, ISO/IEC 42001 and FRAI. Runs in the browser, BYOK, deployed on GitHub Pages.
An open-source CV builder for AI engineers. One brain, three doors: web, CLI, and a Claude skill.
10 Days of LangChain 2026 is a hands-on course that walks you step-by-step through building a fully agentic AI Learning Assistant using LangChain, LangGraph, and modern RAG techniques, from core agent logic to a complete Streamlit app.
A production-style FastAPI NLP service demonstrating async financial text preprocessing with transformer sentiment analysis, spaCy tokenization & entity extraction, rate limiting, and SQLModel persistence, designed to mirror real-world financial AI pipelines.
An experimental GPT-style NLP repository combining data extraction, bigram language modeling, custom training scripts, and early chatbot prototypes, used to explore end-to-end LLM concepts from dataset preparation to inference.
AI-powered chatbot system for dental appointment scheduling - Teraleads Assessment
Past clients & companies
Selected companies and clients I've built with — across enterprise fintech, AI startups and consulting.
Past sponsors
Brands that have backed @sebuzdugan content.