Elaine Bao - Software Engineer at intel

Elaine Bao

Software Engineer at intel

Shanghai, China

Join Prog.AI to see contacts

Summary

🤩

Rockstar

Elaine Bao is a software engineer based in Shanghai with 10 years of experience specializing in machine learning systems and high-performance back-end development. At Intel she focuses on ML engineering and has made substantive open-source contributions to high-profile projects like oneDNN and Apache MXNet. Her work centers on graph backends, ConvTranspose support, format canonicalization, and quantized (int8/uint8) batch normalization optimizations—areas that improve both correctness and runtime performance for production ML workloads. Elaine’s contributions span cross-file schema and shape-inference changes, backend transformation passes, and unit tests, demonstrating a blend of deep algorithmic understanding and pragmatic engineering. She brings an engineer’s attention to data layouts and inference pipelines, often tackling non-obvious format and permutation issues that unlock hardware efficiency.

10 years of coding experience

Github Skills (21)

c-language10

edn10

batch-normalization10

mxnet10

deep-learning10

convolution10

quantization10

cpp10

cprogramming-language10

graph-theory10

operation9

performance-monitor9

tensorrt9

tensorflow9

performance-analysis9

Programming languages (4)

C++CHTMLPython

Github contributions (5)

uxlfoundation/oneDNN

Dec 2020 - Jan 2023

oneAPI Deep Neural Network Library (oneDNN)

Role in this project:

Back-end Developer

Contributions:130 reviews, 189 commits, 17 PRs in 2 years

Contributions summary:Elaine primarily contributed to the graph backend of the oneDNN library, specifically focusing on supporting ConvTranspose operations. They introduced support for different weight formats (IOX/XOI) in the graph interface and backend, involving modifications across multiple files including op_schema.hpp, logical_tensor.hpp, op_def.hpp, and shape_infer.cpp. The user also implemented related changes in the backend dnnl transformation passes, including canonicalization and insertion of permute operations to handle different data formats, along with unit tests for the ConvTranspose functionality.

bfloat16sse41avx512openmpopencl

apache/mxnet

Aug 2019 - Apr 2020

Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more

Role in this project:

Back-end Developer

Contributions:10 commits, 14 PRs, 59 comments in 8 months

Contributions summary:Elaine primarily contributed to the implementation and testing of int8 and uint8 batch normalization (BN) within the MKLDNN backend for the MXNet deep learning framework. This involved adding new implementations, updating existing ones, fixing bugs, and ensuring compatibility with various configurations and hardware. The user also addressed issues related to the use of global statistics and format handling within the MKLDNN BN implementation, and also added support for RROIAlign operator. The work shows a focus on performance optimization and supporting quantized deep learning models.

pythonschedulerdataflowmutationdata-science

Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.

Request Free Trial