Machine Learning Engineer & Researcher Bridging the gap between Causal Inference and production-scale AI.
Currently, I work as a Data Scientist at INGKA Group (IKEA), architecting the recommendations infrastructure serving millions of users. My focus is on building reliable, interpretable systems, whether that's deploying Multi-Armed Bandits in Golang or mapping the internal reasoning of LLMs.
- 🏆 Oral Presentation at CLeaR 2025: My paper on Causal Abstraction in Large Language Models was selected for an oral presentation. I study how models reason, not just what they predict.
- 🛠 Production RL at Scale: Developed IKEA's shift to dynamic Multi-Armed Bandit agents. I built the Golang serving layer, powering the recommendation panel.
- 🔎 Offline Simulation: Designed the evaluation framework that allows us to test agent safety and convergence before live deployment.
- Languages: Python, Golang, SQL
- Machine Learning: PyTorch, TensorFlow, Scikit-learn, Reinforcement Learning (Bandits)
- LLMs & GenAI: Hugging Face (Transformers, PEFT), LangChain, OpenAI API
- Causal Inference: DoWhy, CausalML
- Infrastructure: Google Cloud Platform (Vertex AI, BigQuery, Pub/Sub)
-
Combining Causal Models for Accurate NN Abstractions Accepted for Oral Presentation at CLeaR 2025. Investigating the mechanistic interpretability of GPT-2 through causal abstractions.
-
Reproducibility Study: Label-Free Explainability Published reproducibility study on unsupervised model explainability.
-
Romanian Political Discourse Analysis In Progress. Analyzing the discourse of politicians in official meetings using NLP techniques.
When I'm not debugging distributed systems or reading papers, I focus on discipline and technique through Pole Fitness 🤸♀️ and Piano 🎹.
