Foundation Models as Infrastructure
Why the next platform layer is model-shaped, and how to build a reusable ML platform around it.
New York
Jul 18, 2023 · 9:00 AM
All speakers
Speaker
Dr. Valentina Cruz
Head of AI Research · Helix Labs
Foundation ModelsMLPlatform
Valentina heads AI research at Helix Labs, where her team treats foundation models as a reusable infrastructure layer. Her 2017 paper on cloud warehouses remains one of the most-cited in the proceedings.
Talks
Foundations of the Cloud Data Warehouse
The opening keynote that launched the conference: separating storage from compute, for good.
San Jose
Sep 19, 2017 · 1:00 PM
Published papers
Self-Improving Training Pipelines with Active Data Selection
R. Mehta, V. Cruz
AI Systems · 2025 · 18 citations
Cost-Aware Scheduling for LLM Inference Clusters
R. Mehta, V. Cruz
AI Systems · 2025 · 19 citations
Multi-Modal Embedding Pipelines for Unified Search Across Data Assets
V. Cruz, S. Volkov
AI Systems · 2025 · 15 citations
Declarative Feature Engineering Pipelines for Real-Time ML Serving
V. Cruz, B. Andersen
ML Infrastructure · 2024 · 27 citations
Online/Offline Consistency in Real-Time Feature Stores
L. Zhao, V. Cruz, R. Mehta
ML Platforms · 2022 · 118 citations
Foundation Models as a Shared Infrastructure Layer
V. Cruz, A. Okafor
AI Systems · 2021 · 224 citations
Foundations of the Modern Cloud Data Warehouse
M. Bell, A. Okafor, V. Cruz
Analytics · 2017 · 348 citations