Career Profile
Data Scientist building production AI systems with 6+ years of experience. Created LangTree for LLM workflow management. Published research in reinforcement learning. Expertise spans LLMs, NLP, time series, and cloud ML infrastructure.
Experience
accelerate
Created LangTree: https://ac8.ai/langtree - a tool for complex LLM workflow orchestration.
Implemented AI-based solutions (RAG, semantic clustering, document processing, synthetic data generation, LLM-as-a-judge) for various industries.
Executed research in AI for healthcare (voice analysis for detecting cognitive impairment).
Executed time series analysis and prediction, advanced statistical analysis, and created reinforcement learning solutions.
Built and managed on-premises GPU workstation running Debian for ML infrastructure. Deployed local LLMs, VertexAI, AWS, Databricks, HPC.
Built ontologies and data models. Created data ingestion and cleaning pipelines.
Technologies: Python, LangChain, LangGraph, HuggingFace, transformers, BERTopic, torch, PyTorch, sklearn, CatBoost, MLFlow, rllib, Ray, AWS, VertexAI, Databricks, Dask, Prefect, Docker, Apptainer, Kubernetes, AAS, OWL
AVL
Designed and built end-to-end machine learning solutions. Established effort for reinforcement learning resulting in conference publication.
Set up data ingestion pipelines from vehicle to the cloud.
Executed data science projects for leading automotive OEMs.
Designed digital twin software.
Mentored younger colleagues.
Technologies: Python, RLlib, MLFlow, Poetry, Prefect, Dask, Docker, Kubernetes, Helm, Azure, Linux, git
BID
Fine-tuned early Bert model with touristic sector's data. Integrated it with RASA before it was included in RASA's default offer.
Technologies: Python, RASA, HuggingFace
deeplearning.ai
reviewed and participated in improving two courses in NLP specialization.
Technologies: Python, nltk
Bricsys
Created prototype for NLP system for automated processing of support requests.
Technologies: Neo4J, Gensim, NLTK, Jupyter, BeautifulSoup
Ericsson Nikola Tesla
In team of 3 implemented reinforcement learning agent for Ludo game. Also created a game engine + visualization.
Technologies: Python, Keras, git, Tkinter
Ericsson Nikola Tesla
Created prototype for text searching algorithms on Oracle.
Technologies: Java, Oracle, PL/SQL