Research Overview
Developed AI models for real-time RF signal classification on edge devices, focusing on efficient deployment of deep learning models on FPGA hardware.
Technical Work
Temporal Convolutional Networks (TCNs)
- Built TCN architectures using PyTorch for RF signal classification
- Deployed on AMD Versal VCK190 FPGA
- Leveraged model quantization for efficient inference
- Integrated AI/DSP engines to expedite model inference in edge computing
Edge Computing Optimization
Focused on making deep learning models viable for edge deployment:
- Model compression and quantization
- Hardware-aware optimization
- Real-time inference on resource-constrained devices
Leadership
Scrum Lead for 8-member Agile team:
- Facilitated daily stand-ups and sprint planning
- Coordinated with product owner on requirements
- Managed project timeline and deliverables
Impact
Demonstrated that state-of-the-art deep learning models can be deployed on edge devices for real-time signal processing, enabling autonomous decision-making in bandwidth-constrained environments like satellite systems.