CUDE /GPU DSP Engineer

Job Description

Type: Contractor (3–6 months, P2 Priority – Aerial partnership)

The CUDA / GPU DSP Engineer develops high-performance GPU kernels for real-time digital signal processing in the CLEARSITE™ system. The role focuses on telecom physical layer operations (PSS/SSS detection, PDSCH decoding) on NVIDIA Jetson AGX Orin. This is GPU optimization for signal processing, not ML model work.

Key Responsibilities

• Develop CUDA kernels for PSS/SSS detection and PDSCH decoding
• Integrate with NVIDIA cuRAN libraries and optimize GPU-accelerated pipelines
• Migrate x86 DSP pipelines to ARM + CUDA co-processing
• Implement FFT, matched filtering, and channel estimation routines
• Build Aerial SDK data exchange APIs (IQ data, metadata, bidirectional flow)
• Benchmark and validate kernel performance against latency and throughput targets
• Document kernels, optimization decisions, and integration points

Required Qualifications

• 3+ years writing production CUDA kernels
• DSP fundamentals: FFT, filtering, correlation, modulation, channel estimation
• Telecom physical layer experience (LTE / 5G NR)
• C/C++ and CUDA C proficiency
• GPU memory management, occupancy, and warp-level programming

Preferred Qualifications

• NVIDIA Jetson platform (JetPack SDK, ARM+GPU co-optimization)
• SDR pipelines, IQ sample processing
• 5G NR standards familiarity (3GPP 38.211, 38.212)
• cuFFT, cuBLAS, and GPU signal processing primitives
• PPS/GPS synchronized timing