MELIH UNSAL
Hi! I'm Melih.
Applied AI Engineer with 9+ years of experience building production machine learning systems spanning computer vision, large language models, agentic AI, and retrieval-augmented generation (RAG). Experienced designing AI platforms end-to-end—from model training and fine-tuning to scalable backend infrastructure, knowledge graphs, semantic search, and cloud-native deployment. Creator of DemoGPT, an open-source framework with 1.8K+ GitHub stars that simplifies prompt-driven AI application development. Combines deep machine learning expertise with strong software engineering fundamentals to build intelligent systems that are scalable, reliable, and production-ready.
OPEN SOURCE
DemoGPT (⭐1.8K+ GitHub Stars)
Created an open-source framework that enables developers to build prompt-driven AI applications through an intuitive workflow-based approach. The project has attracted more than 1,800 GitHub stars, reflecting strong adoption within the AI developer community and a commitment to making modern LLM application development more accessible.
REVIEWS FROM CLIENTS
Delivered production AI solutions for international clients, designing scalable LLM applications and backend services tailored to ente...
Delivered production AI solutions for international clients, designing scalable LLM applications and backend services tailored to enterprise use cases.
- Developed cloud-native AI applications using OpenAI, Anthropic, and Gemini models.
- Integrated LLM workflows into production backend systems, emphasizing maintainability, API design, and scalable deployment.
- Collaborated with distributed engineering teams to deliver AI-powered features across multiple client engagements.
Technologies: Python • OpenAI • Anthropic • Gemini • Azure • GCP
Architect AI systems that automate knowledge work, content generation, recruitment workflows, and recommendation engines through produ...
Architect AI systems that automate knowledge work, content generation, recruitment workflows, and recommendation engines through production-ready LLM applications.
- Architected a production multi-agent platform that combines web search, web scraping, and LLM reasoning to automate content generation using continuously updated external information.
- Engineered Retrieval-Augmented Generation (RAG) systems powering internal knowledge assistants capable of answering questions over enterprise documentation with source-grounded responses.
- Designed conversational memory architectures separating short-term dialogue context from persistent long-term memory, improving coherence across extended user interactions.
- Implemented hybrid retrieval strategies combining Pinecone vector search, metadata-aware filtering, and Neo4j knowledge graphs to improve contextual reasoning and retrieval accuracy.
- Developed ETL pipelines synchronizing PostgreSQL and Neo4j, ensuring enterprise knowledge graphs remained continuously aligned with operational business data.
- Optimized image generation models using Parameter-Efficient Fine-Tuning (PEFT) and LoRA techniques to improve domain adaptation while minimizing compute requirements.
- Built recommendation systems that cluster users based on behavioral patterns and continuously refine content recommendations using real-time feedback loops.
- Delivered cloud-native AI services deployed across AWS while collaborating closely with engineering and product teams to evolve production AI capabilities.
Technologies: Python • OpenAI • LangChain • LangGraph • Pinecone • Neo4j • PostgreSQL • AWS • PEFT • LoRA