At DataWise, I worked on AlmaVid, a video-learning SaaS platform, building a Python ETL pipeline for usage logs and a constellation-schema data warehouse. I also delivered five Power BI dashboards integrated into the application.
For AlmaVid’s video content, I developed an end-to-end RAG pipeline with automatic transcription, local embeddings, and similarity search in pgvector. I connected it to Llama 3.3 70B via Groq, with predefined actions and open-ended questions.
I also developed a recommendation engine and a dropout-risk prediction model, and exposed the pipeline and models through a FastAPI microservice for the Next.js application. The service was containerized with Docker and deployed through CI/CD.
Earlier, at DataWise, I supported synchronization from Dynamics 365 Business Central and created Power BI production KPI dashboards. My other projects include an insurance pricing system for Assurance BIAT and web applications built with Laravel and .NET.
