The problem this talk solves: Most enterprise AI budgets are wasted on pilots that never ship. This session names the exact failure modes — data pipelines, organisational resistance, missing MLOps — and the decision framework that gets projects to production.

Key takeaways: The 5 checkpoints every enterprise AI project needs before it touches production. How to structure the ML team for deployment velocity. The MLOps minimum viable stack that's actually achievable.

Why Neil: He's shipped AI systems to production across healthcare, sports analytics, and enterprise tech. He's not speaking from theory — he's speaking from scars.

The problem this talk solves: LLMs are being deployed in regulated industries without the audit trails, bias controls, or governance frameworks that regulators will eventually demand. This talk is the early warning system.

Key takeaways: The Responsible AI audit checklist for LLM deployments. How to build bias detection into your ML pipeline. The 3 governance controls that satisfy most regulatory frameworks today.

Why Neil: Neil has published peer-reviewed research on responsible AI and has built credit scoring models under strict fairness constraints — he understands what "auditable" actually means in practice.

The problem this talk solves: ML teams know how to build models. They don't always know how to run them — at scale, reliably, with costs under control. LLMOps is the missing piece between the demo and the dashboard.

Key takeaways: The LLMOps stack that works in 2026 (without vendor lock-in). Prompt versioning, eval pipelines, and cost observability patterns. How to instrument an LLM application for production debugging.

Why Neil: He built and maintains LLMOps pipelines for production systems — including CricketGPT, an LLM that outperformed SOTA models on inference speed and memory efficiency.

The problem this talk solves: Leaders are drowning in GenAI hype and struggling to separate signal from noise. This talk cuts through with a practitioner's take: what's actually delivering ROI, what's still vaporware, and where to put the next dollar.

Key takeaways: The 3 GenAI use cases with proven enterprise ROI. Why RAG is the most underrated enterprise AI pattern. The 5 questions to ask any GenAI vendor before signing.

Why Neil: Neil has deployed production GenAI systems for clients — not just prototypes. He brings honest metrics, not sales slides.

The problem this talk solves: Visual inspection, quality control, and medical imaging are ripe for AI — but most organisations don't know how to build the business case or scope the engineering work. This talk does both.

Key takeaways: How to quantify the ROI of visual AI before writing a line of code. The 3 production patterns for computer vision that are actually reliable. How to present a CV project to a CFO (and win).

Why Neil: He's built thermal imaging systems for breast cancer detection and visual AI for sports analytics — two very different verticals that share the same engineering fundamentals.