Anvaya is an AI platform that turns wafer defect maps, process sensor data, and fab telemetry into real-time yield intelligence — closing the gap between what a fab produces and what it could.
Advanced chip manufacturing involves over 700 sequential process steps, each generating sensor readings, metrology measurements, and defect inspection images. By the time a wafer reaches final test, actionable signals from step 47 are long buried under noise.
Conventional statistical process control (SPC) catches drift after the damage is done. Root-cause analysis is still largely manual — engineers spending weeks correlating defect maps to process excursions. Meanwhile, every failed wafer at an advanced node represents $10,000–$50,000 in sunk cost.
The problem compounds across the supply chain: a yield excursion at one fab disrupts allocation across every downstream customer, with no intelligent rebalancing layer to absorb the shock.
Amit built Anvaya from a single conviction: that semiconductor manufacturing is the most data-rich industry on earth, and also one of the least data-driven in its decision-making. Having spent years at the intersection of process engineering and machine learning, he saw firsthand how yield engineers made billion-dollar decisions from incomplete information — and how much systematic signal was being discarded at every step.
Anvaya is his answer to that gap. Not a dashboarding tool, not another SPC add-on — but a full inference layer that treats the fab as a learning system. Amit's thesis is that the next competitive frontier in semiconductors isn't lithography equipment or EUV access. It's who turns process data into compounding yield advantage faster.
We're working with a select group of fabs and IDMs on early access pilots. If you run yield engineering at an advanced node, we'd like to talk.