Open to opportunities

Iaroslav Igoshev

@iaroslavigoshev

Message

Senior software engineer specializing in HPC, distributed systems, and GPU acceleration.

Germany

Message

What I'm looking for

I seek a technical role where I can lead engineering efforts in HPC/AI, optimize distributed GPU workloads, mentor teams, and drive measurable performance gains in a collaborative, innovation-focused environment.

I am a Senior Software Engineer with 7 years of experience building and optimizing high-performance computing and AI libraries, contributing substantially to open-source projects such as Modin and unidist. I focus on distributed computing frameworks (MPI, Ray, Dask) and GPU acceleration (CUDA, OpenCL) to deliver scalable, production-ready solutions that measurably improve performance.

I have led cross-functional teams, automated release processes, and implemented performance optimizations that produced up to 2x speedups versus pandas and up to 30% improvements in algorithm throughput. I emphasize mentorship, reproducible benchmarking, and pragmatic engineering to reduce manual effort and accelerate adoption of advanced compute technologies.

Experience

Work history, roles, and key accomplishments

ML Engineer

AI Startup

Apr 2025 - Aug 2025 (4 months)

Researched GPU sharing strategies to enable concurrent AI workloads on single GPUs and evaluated technologies (NVIDIA MIG, MPS, HAMi-core, RunAI) and CPU-based time-slicing to identify trade-offs for different workloads.

GPU Sharing NVIDIA MIG RunAI Performance Analysis CUDA Benchmarking Research Nvidia MPS

AI Frameworks Engineer

Intel

Oct 2021 - Mar 2025 (3 years 5 months)

Enhanced Modin with scalable distributed operations and built unidist to unify distributed execution, introduced MPI dynamic process spawn and shared-memory optimizations that improved performance (up to 2x vs pandas on H&M workload; ~10% vs Ray in some scenarios).

Modin Unidist MPI Ray Distributed Computing Python PyPI Automation Performance Optimization CI CD

Machine Learning Engineer

Intel

Oct 2019 - Oct 2021 (2 years)

Implemented an SGD prototype in oneDAL to overlap kernel execution and data transfer achieving up to 30% speedup and developed automated benchmarking with weekly reports to monitor performance and accuracy across algorithms.

OneDAL SGD Performance Engineering Benchmarking Automation)CUDA CPU GPU Overlap Python

Software Validation Engineer

Intel

May 2018 - Oct 2019 (1 year 5 months)

Validated Autonomous Driving Library algorithms by developing and executing extensive test suites (tens to hundreds of tests monthly), debugging, and writing test specifications to ensure timely, high-quality releases.

Test Development Debugging Test Specifications QA Automation Autonomous Linux Performance Testing

Education

Degrees, certifications, and relevant coursework

Lobachevsky State University of Nizhny Novgorod

Master of Science, Information Technology in Physical Research

2014 - 2020

Completed an MSc in Information Technology in Physical Research with coursework in HPC technologies (CUDA and MPI) focusing on cross-ambiguity function of broadband signals.