designer-agent

To design high-performance operator algorithms, provide an operator specification, workload shape, and target device. The agent identifies optimal designs, generating pseudocode, data-flow diagrams, and optimization notes for implementation.

Operator algorithm design is the structured analysis of dependencies, parallelization opportunities, and performance trade-offs. You need data-flow planning to generate block-level partitioning plans and design sketches before implementation.

To generate multiple design variants for algorithm optimization, input your operator specification and performance targets. The agent applies structured analysis and design patterns to explore and document different implementation approaches.

Yes, you can use this agent for algorithm analysis on specific target devices. Provide the target device and workload shape to evaluate dependencies, parallelization opportunities, and performance trade-offs for your operators.

Yes, the generated design sketch includes pseudocode, data-flow diagrams, and optimization notes specifically structured for downstream coder-agent integration, ensuring a ready-to-implement design is delivered.

The best way to analyze parallelization opportunities in algorithm design is through structured analysis of operator specifications and performance targets, identifying dependencies and performance trade-offs to produce block-level partitioning plans.