AI Engineer focused on building LLM-powered applications, RAG systems, and automation workflows.
I work with Python, LangChain, and tools like Claude to turn language models into practical, working products — from document Q&A systems to multi-agent automation pipelines. My background is in software engineering (iOS), and I moved into AI engineering through hands-on projects and continuous learning.
Python · LLMs · RAG · LangChain · LangGraph · n8n · FastAPI · PostgreSQL · Docker · Prompt Engineering
EchoExtract — Video-to-text transcription with faster-whisper Local GPU-based transcription with a clean, layered architecture (Strategy pattern, Pydantic). Optimized for ~2.5× faster inference through batched processing and VAD filtering. Supports 99+ languages including Persian.
chunking-strategies-comparison — Comparing 5 chunking strategies for RAG Implements and benchmarks fixed-size, recursive, token-based, structure-aware, and semantic chunking from scratch — with a focus on Persian text and an honest analysis of the trade-offs.
customer-triage — AI customer message triage with LangChain Classifies incoming messages, extracts structured data, and drafts replies in Persian and English.