ABDULHAMID A. ADEBAYO

Designed and implemented reinforcement learning-guided supervised fine-tuning to customize large language models for complex agentic IT automation use cases.

Developed and deployed continuous pre-training and post-training pipelines, integrating supervised fine-tuning and reinforcement learning, to customize base models for domain-specific applications like unit test generation.

Engineered and maintained robust automation tools, optimizing model training workflows within high-performance GPU environments.

Spearheaded the data curation and pretraining of the Granite 165M model, serving as the foundational base for granite-docling-258M, which achieved superior multilingual document parsing performance against smol-LM benchmarks.

Pioneered novel data processing and filtering methodologies, validated by ablation studies, which boosted engineering efficiency and yielded high-quality tokens for LLM training.

Streamlined the end-to-end data pipeline from acquisition to token delivery, reducing human intervention by up to 80% and accelerating LLM development cycles.

Developed and curated critical pre-training datasets, including GneissWeb, specifically designed for advanced language and code models.

Managed and optimized large-scale CPU clusters, ensuring efficient and reliable data processing operations for LLM development.

Provided strategic support to the VP of AI Research, contributing to the planning and execution of key initiatives, quarterly roadmap development, and cross-functional project alignment.

Served as a central liaison, enhancing communication and collaboration between senior management, research scientists, and product teams to ensure timely achievement of project milestones.

Authored comprehensive technical summaries, progress reports, and executive presentations, effectively translating complex AI research into actionable insights for leadership.

Developed an AI-powered tool that automated the mapping of security requirements to industry regulations and standards, including NIST 800-53, HIPAA, and PCI.

Automated risk assessments for Cloud and Edge computing resources, leveraging frameworks such as NIST 800-37, FAIR, and CIS RAM to enhance security posture.

Engineered advanced tools and proof-of-concepts for proactive threat hunting, integrating technologies like Kestrel, Stix-Shifter, and Stix-Extensions.

Architected and implemented a trusted service mesh, significantly advancing Zero-Trust Security principles and infrastructure.

Led and managed multiple information security management system (ISMS) projects, ensuring strict adherence to ISO 27001 standards.

Conducted comprehensive security audits, including penetration tests and security control assessments, alongside privacy compliance evaluations for GDPR and CCPA.

Developed and delivered engaging security awareness and training programs, enhancing organizational security posture and employee compliance.

Developed robust APIs and automated routines to programmatically manage accounting transactions, data backups, and report generation for a core banking application, enhancing operational efficiency.

Created interactive dashboards and analytical tools, providing critical insights to optimize financial services operations.