sij oguntayo

Architected an end-to-end generative AI platform using Python, FastAPI, and open-source LLMs. Built agentic content generation pipelines with LangGraph and a RAG-based brand knowledge system using LlamaIndex.

Designed and built the full AI system architecture for a real-time video surveillance platform. Developed a multimodal model combining CLIP-based vision encoders with LLMs and optimized for real-time inference with TensorRT and ONNX.

Reduced background removal inference time by 9x and increased clothing taxonomy classification precision by over 500%. Built collaborative filtering and matrix factorization models for a social recommender system and implemented a multi-stage RAG pipeline with LlamaIndex.

Built generative AI models for virtual interior staging using Stable Diffusion. Fine-tuned inpainting and ControlNet pipelines and optimized inference with ONNX and quantization.

Introduced transformer-based document intelligence (LayoutLM, BERT) and Graph Neural Networks, reducing extraction error rate by over 75%. Managed a team of 3 engineers across the ML delivery lifecycle.

Increased search click-through rate by 200% by redesigning the recommendation engine and search ranking algorithm for R Discovery. Scaled the data collection pipeline to index over 200 million research papers.

Progressed from Mentor to Senior Mentor to Technical Lead across ML and Deep Learning Nanodegree programs. Reviewed hundreds of ML project submissions and led a team of mentors.

Co-founded and led technology for a ticketing and RFID payments platform generating over $3M in revenue and scaling to 300K+ users. Grew revenue by over 300% and led an engineering team of 6.

Completed an MSc in Computing & IT from the Open University, UK in 2016.