I’m a CTO and AI/ML engineer focused on taking models from notebook to production, especially large-scale fine-tuning and distributed training. I build practical ML systems spanning data pipelines, training orchestration, reliability, and cost optimisation for GPU-heavy workloads. My work includes PyTorch and DeepSpeed training stacks, checkpointing and recovery patterns for long-running jobs, and turning research prototypes into repeatable deployment pipelines. I also develop AI agents that integrate tool-use and evaluation harnesses to deliver measurable outcomes. I have volunteered at DSF and participated in mentoring (October 2025). My talks emphasise pragmatic architecture choices, what breaks at scale, and concrete patterns attendees can apply immediately.
