Sofya F.

Key project outcomes: • Cut data pipeline failures by 90% (using Apache NiFi, Jolt & Python) • Reduced AI model hallucinations by 60% (with RAG & Mistral-7B) • Sped up database reports 3x (via PostgreSQL query optimization) Graduated with a CS Bachelor degree specializing in AI, 1.5 yrs commercial experience. DETAILS: Data Pipeline Engineering (Python & ETL): Experience with ETL pipelines using Python and Apache NiFi. Managed 50+ flows processing 1B+ rows/month. Key tasks: • Python Automation: Custom async scripts for API orchestration and complex data processing. • JSON Transformation: Advanced mapping of nested structures using Python and Jolt. • Enterprise Integration: Scalable Apache NiFi flows with error handling, retries, and AWS S3 connectivity. Results: Reduced data latency by 60% and pipeline failures by 90%. Applied AI and NLP: Development of RAG systems and LLM integration. • Backend: Async APIs using FastAPI and AsyncPG. • Inference: Running quantized models (Mistral/Llama 3) via llama.cpp to reduce token costs. • Search: Hybrid search implementation (Semantic Embeddings + RapidFuzz keyword matching). • Data Processing: Custom async parsers (Spacy, PyMuPDF) for text extraction from PDF, DOCX, and PPTX. Database & Web Scraping: PostgreSQL schema design and query optimization (200+ tables). • Results: Reduced database costs by 25% and increased reporting speed by 3x. • Scraping: Automation using Selenium and BeautifulSoup (99% success rate, 500k+ pages/month). Portfolio attached below. Contact me to discuss technical details.