Alechenu Iyoko

Architected MeO, a clinical-grade multi-agent system using LangGraph, enabling stateful reasoning, protocol guidance, and personalised metabolic insights. Designed and deployed the full backend infrastructure, including a FastAPI application on AWS EC2, a knowledge base in S3, and a Redis Enterprise Cloud vector store. Implemented the complete MLOps pipeline, from data ingestion and embedding generation to containerization with Docker and secure deployment with a Next.js frontend on Vercel. Solved complex cloud networking and dependency issues, successfully debugging AWS security groups, VPCs, and library incompatibilities to achieve a stable, production-ready MVP.

Developed a real‑time visual grounding pipeline, enabling the system to interpret images, maintain state, and generate context‑aware responses. Designed and implemented a multimodal assistant combining vision, language, and structured reasoning, built on a modular agentic architecture.

Engineered an end-to-end text classification pipeline to predict medical conditions from symptoms, achieving an 83.4% Macro F1-score. Implemented a hybrid model using BioBERT embeddings with a LightGBM classifier and deployed the result as an interactive Streamlit application.