Oscar Andersson
Cambridge, England, United Kingdom
Summary
🤩
RockstarOscar Andersson is a software engineer based in Lund, Sweden with seven years of experience specializing in embedded ML and performance-critical back-end development. He has a strong track record at Arm—starting as an intern and master’s thesis collaborator and now a software engineer—contributing to flagship open-source projects like TensorFlow Lite for Microcontrollers and ARM’s CMSIS-NN. His work focuses on quantized ML models and optimized neural-network kernels for resource-constrained devices, including adding quantization-specific ops and performance-focused numerical routines. Oscar pairs practical systems-level coding with research-led rigor from his MSc in Computer Engineering, and he’s helped ship demos and documentation that make ML accessible on microcontrollers. Colleagues rely on him to reduce footprint and boost inference efficiency in constrained environments, a niche combining low-level optimization with applied ML.
7 years of coding experience
Github Skills (32)
machine-learning10
c-programming10
c1110
math10
n10
maths10
digital-signal-processing10
c1710
cmsis10
tensorflow-lite10
sys10
neural-network10
quantization10
mathlib10
embedded10
Programming languages (3)
Github contributions (5)
ARM-software/CMSIS_5
Apr 2019 - Dec 2022
Role in this project:
Back-end Developer
Back-end DeveloperContributions:88 reviews, 127 commits, 326 PRs in 3 years 8 months
Contributions summary:Oscar primarily contributed to the CMSIS-NN library by implementing core functionality for neural network kernels, specifically focusing on improving performance and adding support for different quantization methods. Their work included implementing functions for saturating high multiply and divide by power of two, as well as supporting new depthwise convolution, elementwise add, and matrix multiplication operations. The contributions also involved refactoring existing code and fixing doxygen warnings.
cortexcortex-artoscmsismicrocontroller
tensorflow/tflite-micro
Jul 2022 - Aug 2022
Infrastructure to enable deployment of ML models to low-power resource-constrained embedded targets (including microcontrollers and digital signal processors).
Role in this project:ML Engineer
ML EngineerContributions:1 review, 5 commits, 4 PRs in 23 days
Contributions summary:Oscar primarily contributed to the `tflite-micro` project, focusing on implementing and refining quantized machine learning models for embedded systems. They addressed bugs related to optional bias data in CMSIS-NN, ensuring the correct API usage. The user also added quantization-specific registrations for pooling and other operations like ADD, SVDF, and MUL, which is likely done to optimize the library size and performance for resource-constrained devices. These changes directly relate to the core functionality of deploying machine learning models on embedded targets.
signalml-modelslow-powerprocessorsdeployment
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.