Ai Engineer
Current🔊 Deep Noise Suppression for low power Embedded systems :Contributed the development of an AI-based Noise Suppression (NS) solution using TensorFlow, focusing on end-to-end data strategies, model design, and deployment on a HIFI DSP chip within a System-on-Chip (SoC) running a Real-Time Operating System (RTOS). Designed and trained robust AI models for noise suppression through advanced audio data augmentation techniques, synthesizing diverse datasets to simulate real-world noise conditions. Optimized the models for performance and efficiency, converting them into highly efficient C code and leveraging Cadence libraries for seamless integration. Successfully ported the optimized models to the DSP chip of the SoC, ensuring real-time noise suppression with minimal power consumption on the RTOS platform.🔒 Cyber Security using AI:Spearheaded projects in Speaker Identification and Deep Fake Detection, enhancing law enforcement capabilities in identifying criminals and combating cyber fraud.Developed an ECAPA TDNN model trained on an Indian-specific dataset, achieving remarkable testing EERs of 1.5%.Contributed to the detection of deepfakes across various media formats, ensuring the integrity of digital content.💡 Cost-Effective Data Collection and Mining Tools:Engineered web applications and automation tools for efficient data collection and mining, accelerating processes by up to 8x.Empowered data mining teams with precise data mining capabilities.
