Levon Ghukasyan
Back End Developer at Tmarket
Yerevan, Armenia
Summary
🤩
Rockstar🎓
Top SchoolLevon Ghukasyan is a back-end developer based in Yerevan, Armenia with three years of professional experience building reliable server-side systems and leading small teams. He has worked across startups and product companies (Tmarket, Cell Annotation Platform, AOD.space) and evolved from full‑stack and team‑lead roles into a backend/DevOps specialist. An active open-source contributor, Levon improved the DeepLake data loader for activeloopai—optimizing NumPy collation, eliminating unnecessary upcasting, and hardening S3 object-size retrieval for more resilient cloud storage. He combines hands-on performance tuning, dependency and release management, and practical experience deploying data-centric workloads for ML pipelines. Colleagues would call him pragmatic and detail-oriented, with a knack for squeezing extra efficiency from data pipelines.
4 years of coding experience
3 years of employment as a software developer
2, web frontend developer, 2, web frontend developer at Yerevan State University
Github Skills (8)
python10
numpy10
pytorch9
dependency-management8
data-science7
vector-search7
machine-learning7
aws-s37
Programming languages (2)
Github contributions (5)
activeloopai/deeplake
Nov 2021 - Mar 2025
Database for AI. Store Vectors, Images, Texts, Videos, etc. Use with LLMs/LangChain. Store, query, version, & visualize any AI data. Stream data in real-time to PyTorch/TensorFlow. https://activeloop.ai
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:108 reviews, 189 PRs, 226 pushes in 3 years 4 months
Contributions summary:Levon primarily focused on improving the `DeepLakeDataLoader` class, specifically optimizing the collation of NumPy arrays, removing unnecessary upcasting operations, and addressing dtype handling within the data loader. Their contributions also included bumping the `libdeeplake` version and merging various branches, indicating involvement in dependency management and release processes. Additionally, the user made adjustments to the S3 provider for more resilient performance, including retrying object size retrieval for improved stability with cloud storage.
pythonreal-time-datadata-streamjupyter-notebookvisualize
activeloopai/buH
Aug 2024 - Mar 2025
Hub backwards (buH) compatibility tests!
Contributions:15 pushes in 6 months
testingcompatibility
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.