Richard O. Ojo

Led the architecture and delivery of enterprise agentic AI solutions connecting large language models to internal knowledge, APIs, operational systems, and human-approval workflows. Built production generative AI applications on Amazon Bedrock with Anthropic Claude and Amazon Nova, enterprise RAG platforms using Bedrock Knowledge Bases, S3, and OpenSearch, and multi-agent workflows with LangGraph and Strands Agents. Implemented AgentCore runtime, gateway, memory, identity, observability, evaluations, guardrails, secure tool access, CI/CD, containerized services, and Python/FastAPI applications.

Led applied AI and machine learning solutions for tax, accounting, finance, and advisory workflows. Built financial ML models and data pipelines for anomaly detection, classification, forecasting, and variance analysis. Developed NLP and document-intelligence pipelines for extraction, classification, entity recognition, semantic similarity, and search. Delivered RAG systems with embeddings and vector search, piloted Amazon Bedrock, created Python/FastAPI/Flask services on AWS, defined evaluation frameworks, and implemented human-review workflows for sensitive financial outputs.

Developed PyTorch computer vision pipelines for image classification, object detection, segmentation, and attribute recognition using CNNs and Detectron2/Faster R-CNN. Delivered visual product-understanding features for Facebook Marketplace using image embeddings, category and attribute prediction, and similarity search. Built image-similarity and near-duplicate matching workflows, created training and evaluation pipelines for large image corpora, and optimized distributed training, inference throughput, memory, and latency.