Few artificial intelligence (AI) areas have garnered as much attention or posed as many challenges as edge computing. Characterized by running AI models on local devices rather than relying solely on cloud infrastructure, edge computing demands unprecedented efficiency, privacy, and adaptability. Param Popat, a machine learning engineer at Apple, is at the center of this transformative field as he transforms what is possible in AI.
Param Popat's innovations range from pioneering photorealistic 3D simulations to advancing AI security protocols, all while tackling the limitations of edge computing. "I have always believed in pushing the boundaries of what AI can achieve while ensuring it remains accessible and secure," he reflects.
A Career Forged in Academia
Param Popat's foundation in AI was laid at Columbia University, where he completed a Master of Science in Computer Science with a near-perfect GPA. His academic pursuits were not confined to theoretical research. During his time there, he authored a paper on using deep learning for object identification on edge devices like Raspberry Pi. "The goal was to make AI usable in environments with limited resources, which remains a cornerstone of my work today," he explains.
After completing his education, Popat embarked on a career path that saw him tackle some of the most complex challenges in AI. His internship at Bosch in 2019 was particularly crucial, where he developed a system to protect machine learning models from adversarial attacks, work that later contributed to the establishment of Bosch's AIShield, a subsidiary focused on AI security.
Transforming the Field
At Apple, Param led the development of a 3D photorealistic scene reconstruction pipeline using Gaussian Splatting. He has enabled the creation of simulation environments that replicate real-world conditions with astonishing fidelity. This project has implications not just for product testing but also for the broader application of AI systems. Param Popat helped design AI agents employing advanced reinforcement learning techniques to improve multi-agent interaction performance and safety.
Another of Param Popat's notable contributions is the gesture recognition system for Apple Watch's Double Tap feature. The feature, which allows users to control the device with a simple tap, showcases his ability to blend high-level AI concepts with user-centric design. "The challenge was to deliver real-time performance with minimal battery usage, a requirement that pushed the limits of edge computing," he recalls.
The Role of Edge Computing in AI's Future
Edge computing, the process of running AI models directly on devices, is becoming increasingly critical as AI applications expand into healthcare, consumer electronics, and AI agents . Industry analysts project the global AI market to grow to $826.7 billion by 2030, with edge computing playing a key role in this expansion.
Param Popat has demonstrated how on-device AI can meet the dual demands of performance and efficiency by optimizing neural networks for Apple Silicon. However, his contributions also highlight the challenges inherent in this field. "Edge computing forces us to rethink everything, from how we design models to ensure data privacy," he says.
Addressing AI's Security Challenges
One of Param Popat's most impactful achievements lies in AI security. His pending patent, "A Method to Prevent Capturing of Models in an Artificial Intelligence-Based System," addresses a critical vulnerability in deploying AI systems. As companies increasingly integrate AI into their operations, protecting intellectual property has become a pressing concern.
"AI is only as strong as its weakest link, and security is often that link," he notes. His work has drawn attention to the need for robust safeguards, particularly as AI models become more pervasive in sensitive industries.
Param Popat's contributions have earned him well-deserved recognition. He was invited to serve as a reviewer for prestigious conferences like ICCV 2023 and Apple's Machine Learning Summit 2025. These roles, he says, are an opportunity to shape the future of AI by fostering rigorous research.
Beyond his professional accolades, Param Popat is deeply committed to mentorship and collaboration. His work on conversational AI during an internship at AI Zwei exemplifies this spirit, as does his development of Quick-ML, an API-based tool for rapid machine learning prototyping.
As AI continues to evolve, so does the debate over its ethical implications. Param Popat's approach, rooted in accessibility, efficiency, and security, offers a blueprint for overcoming these challenges. Yet, he acknowledges that the road ahead is fraught with complexity. His emphasis on privacy-preserving systems is particularly relevant as AI technologies expand into healthcare and financial services. "The real test of AI will be its ability to serve humanity without compromising privacy or trust," he concludes.
