A collection of my presentations and keynotes.
Zurich, Switzerland – Overview of the Allen Telescope Array's next-generation signal processing pipeline, built on NVIDIA Holoscan SDK and NEXUS orchestration for real-time processing.
San Jose, CA – Discussion of the SETI Institute's use of NVIDIA Holoscan for real-time AI-powered technosignature detection of Fast Radio Bursts at the edge.
Milan, Italy – Presented at the 75th International Astronautical Congress, this paper describes BLADE's architecture and how it processes ~1 Tbps of data in real-time for the Allen Telescope Array.
Knoxville, TN – Demonstration of combining BLADE with NVIDIA Holoscan SDK to create a next-generation pipeline using RDMA for GPU transfers and ML-based algorithms alongside traditional DSP.
Knoxville, TN – Update on CyberEther featuring memory interoperability between CUDA, Vulkan, and Metal, a new headless remote interface, and plans for GNU Radio integration.
Las Vegas, NV – A look at the Allen Telescope Array's signal processing chain, demonstrating real-time reception of Mars orbiters and weather satellites using BLADE and CyberEther.
Austin, TX – Joint talk with NVIDIA's Adam Thompson on building scalable real-time sensor pipelines, showcasing Holoscan SDK integration into the Allen Telescope Array's signal processing chain.
San Jose, CA – Deployment of a real-time AI pipeline at the Allen Telescope Array using Holoscan SDK.
Tempe, AZ – Progress on CyberEther's "Heterogeneous First" design, with native support for Apple Silicon (Metal) and Linux (Vulkan), running on devices from iPhones to Raspberry Pis.
Tempe, AZ – How GPU-enabled software-defined telescopes process ~1 Tbps in real-time, with focus on BLADE's modular interface that automatically applies optimizations like CUDA Graphs and JIT compilation.
Aspen, CO – Deep dive into BLADE, the C++20 GPU-based beamformer for the Allen Telescope Array, covering real-time processing of ~1 Tbps and automatic optimizations like CUDA Graphs and JIT compilation.
Seattle, WA – Presentation on BLADE, the GPU-accelerated beamformer for the Allen Telescope Array that combines signals from 42 antennas in real-time for technosignature searches.
Washington, DC – Introduction to CyberEther's GPU-accelerated frequency sink interface with native support for Metal, Vulkan, and WebGPU, adapting to each target device's best-supported backend.
Austin, TX – Techniques for real-time FM demodulation using NumPy and SciPy, with GPU acceleration via CuPy and cuSignal.
Austin, TX – How we ported TurboSETI from CPU to GPU using CuPy, Numba, and custom CUDA kernels to detect narrowband drifting signals in terabits-per-second telescope data.
Virtual – Talk about Software Defined Radio, ARM single board computers, and PiSDR.
Las Vegas, NV – Development of Open Satellite Project software to demodulate and decode X-Band datalinks from NASA and NOAA's next-generation weather satellites.
São Paulo, Brazil – Joint talk with Lucas Teske on the Open Satellite Project software stack for decoding weather satellite data from various orbits.
São Paulo, Brazil – Introduction to Open Satellite Project software for geostationary and polar-orbiting weather satellites.
