Partner Software Engineer · Microsoft Azure

Building intelligent systems for cloud infrastructure at scale.

I work at the intersection of machine learning, distributed systems, and large-scale cloud infrastructure. I founded and scaled Resource Central, Azure's intelligent resource management platform — delivering tens of millions of dollars in annual efficiency gains.

Eli Cortez

About

I'm a Partner Software Engineer at Microsoft, currently in the Office of the Azure Core CTO, working on AI-native engineering systems and large-scale infrastructure optimization.

For nine years I founded and led Resource Central — Azure Compute's intelligent resource management platform — growing it from concept to a production prediction-serving system handling millions of real-time requests across Azure's global fleet. The work was published at SOSP and featured in Communications of the ACM.

Before moving to the U.S., I co-founded Neemu, an e-commerce technology startup later acquired by Linx. I hold a Ph.D. in Computer Science from the Federal University of Amazonas, Brazil, where my thesis was named the country's best CS doctoral thesis by the Brazilian Computer Society.

Featured Work

ACM ASPLOS 2025

Coach

Exploiting temporal patterns to safely oversubscribe all resources across cloud platforms — driving large efficiency gains at fleet scale.

MLSys 2023 Best Paper

VM Allocation with Lifetime Predictions

Learning VM lifetimes to pack the fleet more tightly and reduce stranded capacity — recognized with the MLSys 2023 Best Paper Award.

ACM SOSP 2017

Resource Central

Workload prediction and resource management across Azure's global fleet — the system I founded and led, powering efficiency decisions at hyperscale.

CACM 2020

Toward ML-Centric Cloud Platforms

A vision for rethinking cloud infrastructure around machine learning — from resource management to operational decision-making.

Experience

Partner Software Engineer2025 — present
Microsoft · Office of the Azure Core CTO

AI-native engineering systems, infrastructure optimization, and large-scale systems strategy across Azure.

Founder & Technical Lead, Resource Central2016 — 2025
Microsoft · Azure Compute

Founded and led Azure's intelligent resource management platform — workload prediction, VM lifetime forecasting, and resource allocation across the global fleet — delivering tens of millions of dollars in annual efficiency gains. Published at SOSP 2017 and CACM 2020.

Co-Founder & Chief Scientist2009 — 2013
Neemu E-Commerce Technologies · acquired by Linx

Co-founded a technology startup, growing it from founding team to 100+ employees. Built production ML systems for web-scale crawling, information extraction, and classification.

Selected Publications

ASPLOS 2025
Coach: Exploiting Temporal Patterns for All-Resource Oversubscription in Cloud Platforms
ACM ASPLOS · doi
MLSys 2025
ProtoRAIL: A Risk-cognizant Imitation Agent for Adaptive vCPU Oversubscription in the Cloud
MLSys · paper
SC 2025
Workload Intelligence: Workload-Aware IaaS Abstraction for Cloud Efficiency
ACM/IEEE Supercomputing · doi
ASPLOS 2023
Snape: Reliable and Low-Cost Computing with a Mixture of Spot and On-Demand VMs
ACM ASPLOS · doi
MLSys 2023
Virtual Machine Allocation with Lifetime Predictions Best Paper
MLSys · paper
CACM 2020
Toward ML-Centric Cloud Platforms
Communications of the ACM · pdf · talk
SOSP 2017
Resource Central: Understanding and Predicting Workloads for Improved Resource Management in Large Cloud Platforms
ACM Symposium on Operating Systems Principles · pdf · dataset
More publications
DSN 2023
How Different Are the Cloud Workloads? Characterizing Large-Scale Private and Public Cloud Workloads
IEEE/IFIP DSN · doi
WWW 2022
Spot Virtual Machine Eviction Prediction in Microsoft Cloud
The Web Conference (Companion) · doi
ICDE 2016
ICE: Managing Cold State for Big Data Applications
IEEE Int'l Conference on Data Engineering · pdf
PVLDB 2015
Annotating Database Schemas to Help Enterprise Search
Very Large Data Bases · pdf
SIGMOD 2011
Joint Unsupervised Structure Discovery and Information Extraction
ACM SIGMOD · pdf
SIGMOD 2010
ONDUX: On-Demand Unsupervised Learning for Information Extraction
ACM SIGMOD · pdf
PVLDB 2011
A Probabilistic Approach for Automatically Filling Form-Based Web Interfaces
Very Large Data Bases · pdf
JASIST 2011
Lightweight Methods for Large-Scale Product Categorization
JASIST · pdf
JCDL 2011
Building a Research Social Network from an Individual Perspective
ACM/IEEE JCDL · pdf
WWW 2009
Automatically Filling Form-Based Web Interfaces with Free Text Inputs
Int'l World Wide Web Conference · pdf
JASIST 2009
A Flexible Approach for Extracting Metadata from Bibliographic Citations
JASIST · pdf
JCDL 2007
FLUX-CiM: Flexible Unsupervised Extraction of Citation Metadata
ACM/IEEE JCDL · pdf
WIDM 2007
FleDEx: Flexible Data Exchange
Workshop on Web Information and Data Management · pdf
SBBD 2006
Data and Metadata Extraction in Semi-Structured Text using HMMs
Brazilian Symposium on Databases · pdf

Full list on Google Scholar.

Awards & Honors

Best Paper Award MLSys 2023 — Virtual Machine Allocation with Lifetime Predictions
Best CS Ph.D. Thesis in Brazil Brazilian Computer Society (SBC), 2013
CAPES Thesis Honorable Mention Top 3 nationally, Ministry of Education, 2013
ACM SIGMOD Undergraduate Award 2008
Best Paper & Best Tool Awards Brazilian Symposium on Databases (SBBD), 2009 & 2011