Mobility Engineer
SmartSoc India · Nov 2024 – Present
- Integrated AI/LLM models into edge devices on the iOS platform, enabling on-device intelligence, low-latency inference, and real-time processing without internet connectivity.
- Successfully deployed and optimised Llama-based language models on iPhone devices, enabling fully offline AI experiences with efficient memory and resource utilization.
- Enhanced LLM inference speed by 50% using Metal GPU acceleration and optimized tensor operations on over 10 iPhone models.
- Improved system performance by implementing request batching, GPU-accelerated inference routing, and optimised API communication between device and backend.
- Developed and integrated AI assistants capable of handling natural language queries, content generation, and contextual reasoning directly on-device.
- Implemented model lifecycle management, including model download, caching, versioning, and runtime loading, ensuring reliable deployment and updates of AI models on iOS devices.
- Contributed to the AI Copilot project by designing scalable AI architecture, integrating multiple AI models, and enabling seamless interaction between edge and cloud-based intelligence systems.
- Improved on-device AI processing by integrating LLM models on iOS, supporting offline capabilities with 0% internet reliance.

