Eric Ijeoma

I’m an AI/ML Engineer focused on shipping production-grade LLM applications that are fast, structured, and dependable. At Lincorne Technologies, I built a multi-agent orchestration system that combines a smart LLM brain with live health monitoring so business users get accurate answers to complex questions in under half a second.

I use LangGraph to enforce step-by-step workflows across agents, and I expose the result as a fully stateful multi-agent pipeline via REST APIs with sub-500ms streaming response times. With RAG, NLP, and SQL, I pull the right document sections from structured financial data so the LLM doesn’t guess—reducing hallucinations by 40% and making high-stakes financial decisions more reliable.

I also design secure, scalable deployments using MCP and A2A communication for controlled tool access, with Docker, Kubernetes, and AWS for portability and cloud scaling through DevOps CI/CD. Previously at Midas, I improved in-house AI using LoRA fine-tuning (numpy, PyTorch) for 31% higher accuracy and 60% lower inference cost, and I built permission-based automation across Gmail, Slack, and Notion with zero credential leakage incidents.