Reduce specialty fab excursion diagnostic cycles from days to hours by correlating defect and process data that already exists in the fab.
Wafertune was built by engineers who had spent years inside specialty 200mm fabs watching yield teams do by hand what a correlation model could do in minutes. The founding story is not a pitch about market size. It is a specific week at a specific fab, with a specific defect problem that took four days longer than it needed to.
In mid-2022, Jonas Falk spent six weeks embedded at a 200mm specialty power fab in Phoenix, helping their yield team diagnose a persistent gate oxide defect excursion. He watched three yield engineers spend most of four days manually cross-referencing KLA inspection CSV exports against Lam etch recipe logs in a shared Excel spreadsheet to isolate a single chamber with a marginal RF power drift.
The data needed to close that root cause in under four hours already existed in the fab's own tool data systems. It lived in three different proprietary export formats with no correlation layer. The yield engineering team was doing manually what a pattern-matching model could do in minutes if the data streams were aligned.
Jonas built a Python script that ingested KLA wafer map exports and Lam chamber parameter logs for a single etch step, ran spatial correlation between defect clusters and chamber parameter deviations, and output a ranked hypothesis list to a text file. The fab's yield manager ran it on historical excursion data and matched 11 of 14 past root causes correctly, prompting a request to expand it to all etch and deposition steps.
Wafertune is now a multi-tool defect-to-process correlation engine for specialty 200mm fabs, integrating with KLA, Onto, Lam, Applied Materials, Synopsys Camelot, and Inficon. It delivers ranked root-cause hypothesis cards to yield engineers within hours of excursion detection rather than days.
Reduce specialty fab excursion diagnostic cycles from days to hours by correlating defect and process data that already exists in the fab.
Most specialty 200mm fabs already have the data they need to close excursion events in hours rather than days. KLA wafer map exports, Lam chamber parameter logs, Synopsys Camelot SPC violation exports, and Inficon endpoint traces capture the evidence at the time of the process event. The problem is not data availability - it is data alignment. Wafertune's mission is to close that gap by building the correlation layer that connects what the fab already collects, so yield engineers spend their time on engineering decisions rather than on spreadsheet assembly.
Wafertune is a seed-stage company, currently working with early-revenue customers at specialty 200mm fabs across the US Southwest. We are not a large enterprise - we are a focused engineering team with direct experience in the fab environments and tool ecosystems we serve.
The team was founded by our CEO Jonas Falk (ex-KLA Instruments, 6 years supporting defect inspection deployments at 200mm fabs), Hiroshi Tanaka (ex-Lam Research INSIGHT data platform), and Svetlana Morozova (ex-ON Semiconductor, 7 years in 200mm power device yield engineering). Kweku Asante (MS statistics, Arizona State University) owns the signature library and pattern-matching model. Our combined experience spans the exact tool integrations and yield engineering workflows that Wafertune is built to automate.
Questions about data formats, tool compatibility, or how a pilot would work at your fab? Reach us directly — you get engineers on the call, not a sales sequence.