Alex Guzhva
Principal Software Engineer at Zilliz
Jersey City, New Jersey, United States
Summary
🤩
Rockstar🎓
Top SchoolAlex Guzhva is a Principal Software Engineer based in Jersey City who specializes in high-performance CPU/GPU C++ for similarity search and finance applications. He holds a PhD in Computer Science and an MS in Theoretical and Mathematical Physics from MSU and brings 10 years of experience building low-latency, production systems—from 24/7 trading platforms to cloud-native vector databases. An active open-source contributor, he has improved FAISS and Milvus with AVX2/AVX512 and ARM SVE optimizations to accelerate ANN search at scale. At Meta he tuned services that saved roughly $1M/year in hardware and developed a lossy compression for TB-scale recommendation models, and earlier he implemented custom ANN for faster time-series prediction in quantitative trading. He pairs deep theoretical grounding with hands-on, cross-architecture performance engineering to squeeze maximum throughput from modern CPUs and GPUs.
11 years of coding experience
18 years of employment as a software developer
Lomonosov Moscow State University
English, Russian
Github Skills (19)
similarity-search10
c-language10
vector-database10
vectorization10
avx210
avx51210
go10
performance-optimization10
simd10
c-programming-language10
linear-algebra9
build-automation8
machine-learning8
testing8
vector-search7
Programming languages (6)
Github contributions (5)
facebookresearch/faiss
Mar 2022 - Sep 2022
A library for efficient similarity search and clustering of dense vectors.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:21 reviews, 23 commits, 124 PRs in 6 months
Contributions summary:Alex's commits focus on optimizing the `faiss` library for efficient similarity search, specifically through implementing and improving AVX2 and AVX512 implementations for key functions. Their work includes the creation of AVX2 implementations for `fvec_madd` and improving AVX2 distance computations. They also optimized AVX2 code for IVFPQ scanning and extended AVX2 versions to support polysemous filtering, demonstrating a strong focus on performance improvements in vector similarity search.
k-meansvectorssimilarity-searchmachine-learningsimilarity
milvus-io/milvus
Sep 2023 - Mar 2025
Milvus is a high-performance, cloud-native vector database built for scalable vector ANN search
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:16 reviews, 19 PRs, 143 comments in 1 year 6 months
Contributions summary:Alex primarily focused on improving the performance and stability of the Milvus vector database. Their contributions included fixing a temporary file issue in the testing utilities, removing unused code, and optimizing the bitset utilities. They also worked on performance improvements for bitset operations utilizing AVX512 and ARM SVE, demonstrating a focus on low-level optimization for the core functionalities of the database. Furthermore, the user addressed a compilation issue related to GCC-13 and updated the build process for CMake installation across different platforms.
approximate-nearest-neighbor-searchfaisshnswsimilarity-searchvector
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.