Über Nazarii
Englisch
Muttersprachlich oder zweisprachig
Projekt- und Berufserfahrung
- Software MindJunior AI Engineer – NLP & ML PipelinesDIGITALAGENTUREN & IT-CONSULTINGJuli 2019 - August 2020 (1 Jahr und 1 Monat)Krakow, Polen
- Transformer Models (2019): Experimented with BERT and GPT‑2 for sentiment analysis and text classification tasks, laying the foundation for later LLM work.
- Data Pipeline Construction: Built ETL pipelines using Apache Airflow and Spark to preprocess 10+ TB of multilingual text data.
- Model Deployment: Deployed first production ML model (a fine‑tuned RoBERTa for intent classification) using Docker and AWS ECS, achieving 99.9% uptime.
- Mentorship & Growth: Rapidly advanced from junior to mid‑level engineer within 14 months, receiving “Rising Star” award for outstanding contributions.
- RheinData AIAI Engineer – Generative AI & LLM ProductizationDIGITALAGENTUREN & IT-CONSULTINGSeptember 2020 - April 2022 (1 Jahr und 7 Monate)Cologne, Deutschland
- Generative AI Adoption (2020): Built the company’s first GPT‑3‑powered text summarization and content generation features immediately after OpenAI’s May 2020 beta release.
- LLM Pipeline Development: Created an end‑to‑end data pipeline for fine‑tuning GPT‑Neo and GPT‑J on customer‑specific documents, achieving 85% user adoption within three months.
- Prompt Engineering Framework: Developed a prompt versioning and testing framework that reduced model output variability by 47%.
- Platform Integration: Seamlessly integrated generative AI capabilities into existing microservices architecture using FastAPI and Redis, enabling real‑time inference at scale.
- Nordic AI LabsSenior AI Engineer – RAG & LLM Pipeline SpecialistDIGITALAGENTUREN & IT-CONSULTINGMai 2022 - Januar 2024 (1 Jahr und 8 Monate)Gdansk, Polen
- RAG Architecture (2022): Implemented the Retrieval‑Augmented Generation framework (originally introduced by Lewis et al. in 2020) to ground LLM responses in proprietary company data. Designed a hybrid retrieval pipeline combining FAISS dense retrieval with Elasticsearch BM25, improving answer accuracy from 68% to 94%.
- LLM Fine‑Tuning (2022): Fine‑tuned GPT‑3.5 and LLaMA 2 for domain‑specific question‑answering, achieving sub‑2‑second latency on CPU‑only deployments.
- Gen AI Content Generation: Built a GPT‑4‑based report generation engine that automatically summarizes medical records and legal documents, saving clients 20+ hours per week.
- Evaluation & Monitoring: Established RAGAS and TruLens evaluation pipelines to track hallucination rates, context relevance, and answer faithfulness.
Empfehlungen
Sei die erste Person, die Nazarii empfiehlt
Teile Deine Erfahrung aus der Zusammenarbeit mit diesem Freelancer.
Diese Freelancer passen auch zu Ihren Kriterien
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Ausbildung und Abschlüsse
- B.Sc. in Computer ScienceWarsaw University of Technology2021