Profile banner
Davit Buniatyan

Davit Buniatyan

Founding CEO at Activeloop β€” building Deep Lake

davit
Mountain View, California, United States
Joined January 2025
Favicon imageloqsh.com
Favicon imagegithub.com
Favicon imagescholar.google.com
Favicon imageactiveloop.ai
Favicon imageycombinator.com

Network

10.9K connections
🌐
Senior Tech & Product Leaders
πŸš€
AI Startup Founders
πŸ€–
AI/ML & Data Scientists
πŸ’»
Software Engineers & Architects
πŸ“ˆ
Business Development & Product Leaders
πŸ’°
Venture Capital Investors
πŸ’‘
Builders

Summary

Founder and product-engineering leader focused on data infrastructure for multimodal AI. As Activeloop's founding CEO, Davit has driven the vision and engineering of Deep Lake β€” a dataset/tensor lake and vector-store designed to store, version, stream, and visualize images, video, audio, text, and embeddings for model training and retrieval-augmented workflows. He combines open-source stewardship with enterprise product work and fundraising (YC S18 alum). activeloop+2
Academic researcher in computer vision and connectomics with applied systems work. Davit's research during his PhD work at Princeton (advised by H. Sebastian Seung) produced papers on metric learning, template matching, and efficient 3D ConvNet inference; he held the Gordon Wu Fellowship and an AWS ML research award while working on large-scale, multimodal neuroscience datasets β€” experiences that motivated building better infrastructure for AI data. google+2
Active open-source contributor and technical communicator. Davit maintains a public GitHub account with repositories tied to Activeloop/Deep Lake and frequently authors technical posts and tutorials about using Deep Lake (company blog and docs), signaling hands-on involvement in product implementation, developer experience, and community adoption. github+2
Bridges academia and startup ecosystems. Davit's trajectory (UCL β†’ Princeton research fellowships β†’ Y Combinator) reflects a background of rigorous research applied to practical product problems β€” from connectomics and efficient model inference to building dataset infrastructure used by both research labs and enterprises. mlconf+2

Work

Education

Projects

Writing

Deep lake: A lakehouse for deep learning

January 1, 2022

Paper / technical writeup describing Deep Lake as a lakehouse-style storage and management solution tailored for deep learning datasets and workflows (multimodal, versioned, vector search).

Favicon imagescholar.google.com

Weakly supervised deep metric learning for template matching

January 1, 2019

Research on improving template matching robustness using weak supervision and deep metric learning techniques (conference paper).

Favicon imagescholar.google.com

PZnet: Efficient 3D ConvNet Inference on Manycore CPUs

January 1, 2019

Work on a CPU-only engine for efficient inference of 3D convolutional networks, addressing cost-effective deployment of deep models on manycore CPU infrastructure.

Favicon imagecollaborate.princeton.edu