Vasilis Papaioannou

About Vasilis Papaioannou

12+ years of experience in delivering impactful data-driven solutions across consulting…

Experience

1. Data Scientist Consultant

   Butterflai

   May 2024 — Present

   Designed and implemented a scalable NLP/RAG workflow to extract structured evidence from 52K+ clinical notes (F1 0.85, ECE 0.08, MIMIC-IV/BIDMC) enabling multi-cohort discovery. Integrated Transformer LLMs (OpenAI gpt-4o-mini) with optimized prompt engineering minimizing latency and costs. Provided extended observability via JSON outputs. Validated LLM outputs with XGBoost and regex-based negation handling. Evaluated with scikit-learn (AUC, sensitivity, specificity, precision, F1).• Developed a hybrid semantic retrieval pipeline integrating word embeddings (Sentence-Transformers, all-MiniLM-L6-v2, IVF, HNSW) stored in PostgreSQL (pgvector, pg_trgm), with spaCy, regex, section parsing for note normalization, curation, and LLM preparation. • Engineered modular Python pipelines for cohort discovery with logging, documentation, and multiprocessing accelerating SDLC. Leveraged Docker and Bash for cross-platform deployment. Git.• Prototyped a FastAPI-based controller – worker service with Hugging Face embeddings (facebook/dinov2), similarity search, and caching (in-memory + persistent). Documented testing.• Built ETLs in SQL mapping EHR data to the OMOP CDM facilitating collaboration across teams. Developed predictive ML models in Python/R for hospital course forecasting of 365K+ ICU patients.

Education

• Johns Hopkins Whiting School of Engineering

  Master of Science - MS, Computer Science

• National Technical University of Athens

  Master of Engineering - MEng, Computer Engineering

• National Technical University of Athens

  Bachelor of Engineering - BE, Computer Engineering