Serge Panev
Deep Learning Software Engineer at NVIDIA
- Role
- Deep Learning Software Engineer at NVIDIA
- Location
- Mountain View, CA, US
- LinkedIn followers
- 500 followers
About Serge Panev
Part of the Deep Learning Frameworks team at NVIDIA, currently focusing on the…
Experience
Deep Learning Software Engineer
Oct 2018 — Present · San Francisco, CA, US
Part of the Deep Learning Frameworks Core team.Responsible in the integration of NVIDIA hardware and software technologies into deep learning frameworks internals (backend and frontend). Accelerating both training and inference performance of modern deep learning workloads. Worked on the high performance data loading and data augmentation for deep learning library DALI- Pipeline composed of operator with DAG executor- Operator composed of custom C++ and CUDA kernels - nvJPEG for HW-accelerated JPEG decoding and NVDEC for HW-accelerated video decoding- C++ and Python frond-endOptimizing and training deep learning models, writing tutorials, demo and blogposts about them.Worked as a core framework developer on Apache MXNet, a highly scalable open-source deep learning software framework, used to train, and deploy deep neural networks.My work consisted of- Profiling and benchmarking deep learning applications (training and inference of models) on the framework and the framework itself- Find bottleneck and optimizing critical paths: by having a better understanding of the model and how it interacts with the hardware and by writing optimized kernels- Design and implement high level (Python) and low level features (C++, CUDA), to help the framework use the most of our company\'s hardware- Writing tests and tools related to benchmarking and deployment running and automating them on NVIDIA\'s Deep Learning servers (DGX-1, DGX-2 and newer systems)Working on PyTorch Geometric, a library built upon PyTorch to easily write and train Graph Neural Networks (GNNs) for a wide range of applications related to structured data.My work mainly consists of- Improving and optimizing the GNN training, inference, deployment workflows on NVIDIA systems- Maintaining the PyG container- Intersection of GNN and LLMs
Skills
- Java
- Html
- Script Shell
- Javascript
- Computer Forensics
- Cuda
- Service Client
- Linear Algebra
- Network Security
- Jquery
- Gestion De Projet
- Unix
- Information Security
- X86 Assembly
- C++
- Conception De Produit
- Swift
- Mysql
- Data Science
- C
- Network Administration
- System Administration
- Microsoft Sql Server
- Css
- Asp.net Mvc
- Machine Learning
- C#
- Sql
- Linux
- Postgresql
- Développement De Logiciel
- Git
- Gestion D'équipe
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.