Skyworks Solutions Global Head of AI/ML Mumtaz Vauhkonen discusses AI agents in semiconductor test automation
In a LangChain podcast episode, Skyworks Solutions' Global Head of AI/ML Mumtaz Vauhkonen and AI engineer Kordag Mehmet Kilic detail how they use LangGraph and LangSmith for end-to-end test automation. The discussion highlights a reduction in root cause analysis time from 20 days to 10 on a first pass, alongside insights on moving from homegrown tools to standardized platforms for on-prem and cloud data integration.
Skyworks Solutions' Global Head of AI/ML and Data, Mumtaz Vauhkonen, along with AI engineer Kordag Mehmet Kilic, recently discussed the company's deployment of AI agents in semiconductor workflows. In this LangChain podcast episode, they explain how LangGraph and LangSmith are being used for end-to-end test automation, root cause analysis, and design assistance.
A key metric highlighted is the improvement in root cause analysis efficiency. According to the episode description, a manual process that previously took about 20 days has been reduced to 10 days on a first pass using these agentic systems. Each use case within their workflow is tracked with its own specific Key Performance Indicators (KPIs).
The conversation also covers the transition from homegrown tooling to LangSmith. Kordag Mehmet Kilic explains that this shift allowed deployment, streaming, and thread management to come out of the box, while evaluations could run on their own datasets and metrics. Mumtaz Vauhkonen details why Skyworks chose LangGraph and LangSmith to handle complex environments involving on-premises and cloud data, third-party tools, and multiple databases, emphasizing quick iterations to get agents into production.