For the past seven years, Sourabh’s work has centered on architecting and scaling machine learning systems across the United States and India. As a Senior ML Architect, his focus was on the deep infrastructure layer — optimizing distributed training, cutting latency, and ensuring models held up under massive compute demands.
Recognizing that high-performance software is ultimately at the mercy of the physical hardware it runs on, he identified a critical industry vulnerability: the deployment of incredibly complex AI models on top of hardware ecosystems that operate almost entirely on blind trust.
Sourabh stepped away from standard ML infrastructure to solve for this physical layer by founding Assurd Techlabs. Today, he is replacing trust with cryptographic and physical proof; engineering forensic test rigs and thermal validation systems to certify high-end compute hardware before it ever hits a production environment.