Artur Lesniak
Deep Learning Software Developer at Intel Corporation
Gdańsk, Pomeranian Voivodeship, Poland
Summary
🤩
Rockstar🎓
Top SchoolArtur Lesniak is a Deep Learning Software Developer based in Gdańsk with 10 years of experience building high-performance ML systems, currently contributing to Intel's deep learning efforts. He brings a rare combination of long-term software engineering and operational leadership from earlier co-founder and head-of-IT roles, plus postgraduate training in software engineering. Artur is an active open-source contributor to the industrial-grade PaddlePaddle ecosystem, improving MKLDNN integration and performance-critical passes that fuse matmul/transpose/reshape and add DyGraph cache management. He also optimized recommendation-model training by adding BF16 support to PaddleRec’s word2vec benchmark, demonstrating focus on both low-level performance and practical ML workloads. Known for pragmatic, performance-first solutions, he thrives on squeezing latency and resource gains from production AI stacks.
10 years of coding experience
14 years of employment as a software developer
Postgraduate, software engineering, Postgraduate, software engineering at Politechnika GdańskaMaster of Science - MS, Electronics, Master of Science - MS, Electronics at Politechnika Koszalińska
Github Skills (13)
neural-network10
paddlepaddle10
machine-learning10
deeplearning-ai10
c-language10
deep-learning10
cprogramming-language10
word2vec10
python10
efficientnet10
optimization9
distributed-training9
tensorflow4
Programming languages (2)
Github contributions (5)
PaddlePaddle/Paddle
Apr 2020 - Feb 2022
PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）
Role in this project:
Back-end Developer
Back-end DeveloperContributions:175 reviews, 22 commits, 28 PRs in 1 year 10 months
Contributions summary:Artur's commits focus on enhancing the PaddlePaddle framework, particularly within the context of MKLDNN (Intel's Math Kernel Library for Deep Neural Networks). They implemented a pass for fusing matmul-transpose-reshape operations and added a cache management system for the DyGraph, improving performance. Further contributions include integrating global flags to control the use of MKLDNN and selecting specific MKLDNN operations, enhancing flexibility and debug capabilities.
pytorchpythonparalleldeep-learningpaddlepaddle
PaddlePaddle/PaddleRec
Jul 2021 - Aug 2021
Recommendation Algorithm大规模推荐算法库，包含推荐系统经典及最新算法LR、Wide&Deep、DSSM、TDM、MIND、Word2Vec、Bert4Rec、DeepWalk、SSR、AITM，DSIN，SIGN，IPREC、GRU4Rec、Youtube_dnn、NCF、GNN、FM、FFM、DeepFM、DCN、DIN、DIEN、DLRM、MMOE、PLE、ESMM、ESCMM, MAML、xDeepFM、DeepFEFM、NFM、AFM、RALM、DMR、GateNet、NAML、DIFM、Deep Crossing、PNN、BST、AutoInt、FGCNN、FLEN、Fibinet、ListWise、DeepRec、ENSFM，TiSAS，AutoFIS等，包含经典推荐系统数据集criteo 、movielens等
Role in this project:ML Engineer
ML EngineerContributions:5 commits, 2 PRs, 2 comments in 29 days
Contributions summary:Artur primarily contributed to the `word2vec` benchmark within the PaddleRec repository. Their work involved adding BF16 support, which suggests optimization efforts for model training. Further commits show style formatting and the fix for a missing dependency related to the BF16 implementation. The user is likely involved in improving the efficiency and performance of recommendation model training, specifically focusing on the `word2vec` algorithm.
signdsindcnfibinetyoutube
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.