The Spyre AI Accelerator card empowers enterprises to use multiple AI models for complex business solutions that need more than one level of predictive analysis.
Scroll ↓
IBM
2023 - Present
Because we were introducing a net new piece of hardware infrastructure, our team made it easier for system engineers to set up the cards, and to immediately see the value it added to their AI workloads. This increased customer adoption by reducing the card configuration steps and empowering their telemetry ecosystem with relevant card data about capacity planning, AI workload management, card efficiency, and total cost of operation.
Business impact
90% server purchases include Spyre AI cards
Increase in IBM z17 and Software sales where Spyre hardware can support multiple-model orchestration for agentic AI and code transformation applications (watsonx suite).
User impact
↓ Fewer legacy skills required
Net new configuration steps for Spyre were automated through industry-standard Ansible scripts, decreasing the amount of legacy skills required for early-tenure system engineers.
Contribution
Design Lead
I lead the design strategy and research for end-to-end user experiences spanning card installation, firmware configuration, model deployment, and hardware telemetry.
Spyre AI Accelerator card exploded rendering
IBM Blue Studio
