Leeon JohnAI Engineer
I build complex AI systems with a focus on multi-agent inference engines, reliable RAG pipelines, and automated LLM evaluation. Turning research into production-grade systems.
Featured Systems
OrionAI
A robust multi-agent system orchestrating Planner, Retriever, Executor, and Critic agents. Features Redis-backed memory, structured JSON logging, and autonomous retry logic for complex pipeline execution.
EvalForge (Evalynx)
Production-grade evaluation engine using LLM-as-judge. It scores, compares, and iteratively refines model outputs to ensure reliable AI system behavior and mitigates hallucination.
More Selected Works
Fraud Detection System
Real-time AI fraud detection system featuring a live transaction streaming simulator and an interactive Streamlit dashboard.
View RepositoryEduRAG
Personalized AI learning companion leveraging FAISS vector retrieval, automated text chunking, and LLM-powered tutoring.
View RepositoryCode Insights AI
Intelligent code reviewer combining Python AST-based static analysis with Google Gemini reasoning to detect vulnerabilities.
View RepositoryTechnical Core
The tools I use to build scalable AI systems.