Oluwaseun Adeniji

Developed an LLM-based HR and People application using LangChain and Azure OpenAI, streamlining employee inquiries and improving response accuracy related to policy questions. Built high-performance Java and Python backend microservices for real-time AI inference and Kafka-based data streaming, enabling scalable marketplace analytics across ticket intelligence platforms. Fine-tuned Vision-Language Models with HuggingFace on Azure VMs, achieving a 3% improvement in document form and table extraction accuracy. Implemented forecasting models with XGBoost and LightGBM to predict ticket price movements, inventory trends, and demand fluctuations. Optimized AI services with ONNX to cut production inference time from 6 seconds to 2 seconds.

Developed document question-and-answer system using OpenAI API and LangChain; accelerated search by up to 70%. Built distributed Java Spring Boot and Python backend services supporting distributed ML pipelines, REST APIs, vector search workflows, and scalable AI-powered document intelligence on AWS and Google Cloud. Architected and deployed production ML pipelines using PyTorch, TensorFlow, HuggingFace, MLflow, LangChain, and LLamaIndex, improving system performance by 30%. Streamlined data labeling with Active Learning and Clustering, reducing data labeling duration by 30%. Owned the AI initiatives roadmap, driving experimentation with RAG systems and deep learning; ensured best practices across ML lifecycle and built distributed systems with BigQuery, Pinecone, and MLOps tooling, reducing deployment time by 40% and achieving 99.9% uptime.

Architected Recommendation Engine systems using Python, Scikit-learn, NumPy, and Pandas; integrated with product platforms to drive 30% revenue growth through personalized user experiences. Implemented REST APIs for ML model serving and automated data workflows using SQL and BigQuery; optimized query performance leading to 45% faster inference times. Partnered with product managers to translate AI strategy into functional solutions and delivered iterative improvements through continuous experimentation. Built and deployed Computer Vision models for real-time analysis, improving user engagement metrics by 40% through enhanced visual search capabilities.