Vlad Rastvorov
End-to-End AI Engineer | LLM apps, RAG & Automation
New to Contra
Vlad is building their profile!
Built a machine learning forecasting system that transforms raw hourly weather observations into sequences of analytical ODE segments, then trains a transformer encoder to predict future segment sequences and decode them into hourly forecasts.
The system uses interpretable compressed representations instead of raw time-series values, with segmentation, joint boundary unification, transformer training, ensemble inference and forecast evaluation against persistence, climatology, ECMWF and GFS baselines.
1
6
Designed a private AI knowledge assistant concept for business documents, using document ingestion, text extraction, chunking, embeddings, retrieval-augmented generation and backend API integration to answer questions from internal knowledge sources.
0
11
Built an end-to-end NLP system that converts English text into contract-compliant linguistic JSON for language learning applications.
The pipeline combines spaCy parsing, deterministic skeleton generation, TAM enrichment, local T5-based note generation, multilingual translation enrichment, schema validation and PostgreSQL persistence.
My focus was on production-style AI engineering: structured data contracts, dataset preparation, model integration, local inference, validation, backend persistence and maintainable pipeline design.
1
21