Building agentic AI systems that combine multi-agent orchestration with multimodal intelligence.
I develop LLM pipelines, vision-language systems, and production-grade research tools that solve complex, real-world problems through deep learning, semantic retrieval, and full-stack engineering.
Meet me
Hi, I'm Arman Radmanesh.
I'm a Graduate Research Assistant at the University of Oklahoma, pursuing an MS in Data Science & Analytics. I build agentic AI systems, multimodal LLM pipelines, and production-grade research tools.
My work blends deep learning, computer vision, semantic retrieval, evaluation science, and full-stack engineering to solve complex, real-world problems. I focus on multi-agent systems, LLM engineering, and multimodal ML.
I develop multi-agent LLM systems, RL-based post-training loops, and multimodal evaluation frameworks. My research includes A2A orchestration, vision-language pipelines, and semantic retrieval systems using LlamaIndex, LangChain, LangGraph, and Qwen.
Featured projects
Tech Stack
AI & ML
Backend & Infrastructure
Frontend
Databases
Blog posts
Nov 12, 2025 • 3 mins read
Agentifying Design2Code with AgentBeats
A two-month update on my AgentX project for the Berkeley Agentic AI MOOC, building an agentic test harness around the Design2Code visual-to-code pipeline.
Feb 25, 2025 • 3 mins read
Writing a JavaScript Module That Works in Both CommonJS and ESM
A guide on how to write a module that works in both CoommonJS and ESM
Lab experiments
Conway's Game of Life
Interactive cellular automaton with design and simulation modes. Create patterns and watch them evolve.
Emotion-Reactive Avatar
AI-powered emotional avatar that reacts to your messages with facial expressions and speech.
Prime Patterns Explorer
Interactive Ulam spiral visualization with zoom, pan, and prime pattern coloring modes.