Meelad Dashti

About Me

I’m an AI/ML Engineer currently working at Zirak srl in Turin, Italy, where I architect and develop production-grade AI systems. I specialize in building real-time transcription pipelines, RAG systems, and LLM-powered applications that solve real-world problems.

Currently pursuing my Master’s thesis at the University of Twente on Hyperbolic Compositionality in Vision-Language Models, while completing my M.Sc. in Computer Engineering (AI & Data Analytics) at Politecnico di Torino.

What I Do

My work focuses on cutting-edge AI technologies:

  • Real-time AI Systems: Built a meeting platform with <2s latency using WhisperX and Faster-Whisper
  • RAG & Vector Search: End-to-end pipelines with ChromaDB, semantic search, and cross-encoder reranking
  • LLM Applications: Agentic workflows with LangGraph, fine-tuning LLaMA models with LoRA/QLoRA
  • Multimodal AI: Research in vision-language models and hyperbolic embeddings

Research & Publications

  • Master Thesis (Ongoing): Hyperbolic Compositionality in Vision-Language Models at University of Twente
  • PROFES 2025: “Enhancing Software Maintainability through LLM-Assisted Code Refactoring” (Accepted)

Background

I started as a software engineer working in Android and web development, but my passion for AI led me to specialize in machine learning, NLP, and computer vision. I enjoy the challenge of taking research concepts and turning them into production systems that people can use.


Open to: Industry opportunities, research collaborations, and PhD positions in AI/ML.