Sam Medrington

I design human-in-the-loop AI systems for safety-critical environments where complexity, consequence, and uncertainty intersect.

Design Lead for Cognite Flows - a platform for building and deploying industrial AI applications and workflows grounded in real-world operational data.

With a background in Psychology, Human Factors & Ergonomics, and Computer Science, I have spent the last 20 years designing systems where data, decision-making, and human judgment intersect. Previously Senior Director of Product Design at Cognite and Global Head of UX at BAE Systems Digital Intelligence.

Human-AI systems designApplied research at scalePrototyping in codeHuman Factors & ErgonomicsDecision-support systemsPlatform & ecosystem design

I work by building. Research sets the direction, but many important interaction decisions emerge through prototyping, testing, and iteration in code. AI tools compress the distance between field research and working software.

At Cognite, I led design for Cognite Flows - building the reference applications that established the platform's interaction patterns and quality bar, and defining the certification standards used to review applications built by customers and partners.

My work sits at the boundary between probabilistic model output and human operational judgment - deciding what should be surfaced, suppressed, escalated, or acted upon. Read more →

Production systems shipped Concept to factory floor in 2 weeks Industrial AI workflows deployed with real operational users

Selected work

NVIDIA GTC 2026 session slide - Transforming Industrial Ops: Real-Time Forecasting and Anomaly Detection With Foundational Time-Series Models at Scale

Human Judgment in Foundation Model Anomaly Detection

Designed the human review system between a time series foundation model and the engineers who act on its output.

Demoed live at NVIDIA GTC 2026

Production Flow application in use on the factory floor

Production Flow & Maintenance Support

Led ethnographic research and designed production flow applications for pharmaceutical manufacturing operations at Cognite.

"What the Atlas AI team did in 2 days took our UX team 18 months"

Data Quality governance interface showing validation rules and violations

Data Quality Governance for Industrial AI

Led a co-innovation hackathon to define the governance architecture for AI-generated validation rules in safety-critical infrastructure data. Facilitated discovery, then built the working prototype in two days.

From interface to governance

Cognite Data Fusion Module Map showing data flow connections

Redesigning Data Operations

Led a redesign of Cognite's data management experience, using ethnographic research and AI-assisted prototyping to make complex data operations legible.

Validated in days, not months

Vodafone Permissions & Preferences information architecture site map

Selected Earlier Work

Programmes spanning 15 years - including military decision support under security clearance, GDPR consent architecture for Vodafone, a £10M platform transformation for Lloyd's Register, and investigative workflows for UK law enforcement.

High-consequence environments across domains