Kiryong Ha (하기룡)

About

I am a Principal Engineer (E8) at Meta currently working on hyperscale capacity management and capacity fulfillment for Meta's private cloud.

My work focuses on turning demand forecasts, measured supply, service constraints, and infrastructure availability into reliable capacity decisions. The goal is straightforward to state and hard to execute at scale: put the right compute in the right place at the right time, while preserving reliability and efficiency.

Over the course of my career, I have worked across global traffic routing, throughput autoscaling, live traffic load testing, regional capacity planning, mobile–cloud systems, cloudlets, and edge computing. I received my Ph.D. in Electrical & Computer Engineering from Carnegie Mellon University.

Current Focus

Capacity fulfillment

Translating infrastructure demand into executable plans and making capacity available where services need it.

Hyperscale capacity infrastructure

Building private-cloud capacity management and fulfillment systems for Meta products such as Facebook and Instagram.

Closed-loop systems

Connecting forecasts, measured throughput, placement, traffic, and delivery into reliable operating feedback loops.

Career

Current

Meta

Principal Engineer (E8)

Private-cloud capacity management and capacity fulfillment for hyper-scale products like Facebook and Instagram.

2011–2016

Carnegie Mellon University

Ph.D., Electrical & Computer Engineering

Systems infrastructure for mobile–cloud convergence under Mahadev Satyanarayanan.

Summer 2014

Microsoft Research

Research Intern · Edge Computing

GPU-state migration between edge and data-center systems.

2007–2011

ETRI

Research Staff

Virtualization, on-demand computing, and context-aware healthcare systems.

2000–2007

KAIST

B.S., EECS · M.S., BioSystems

Computing systems, signals, and applied research.

Selected Public Work

2023

Global Capacity Management with Flux

Acknowledged Flux team member in the OSDI ’23 paper

Flux automates regional capacity planning across thousands of services, tens of regions, and millions of servers in Meta’s private cloud.

OSDI ’23 paper
2022

Live Traffic Load Testing at Facebook Scale

Speaker with Lin Xiao · Systems @Scale

A production approach for measuring maximum safe service throughput while workloads, software, hardware, and dependencies continue to change.

Watch the talk
2020

Throughput Autoscaling for Facebook.com

Co-author · Engineering at Meta

Workload-driven sizing combines demand prediction, disaster modeling, live-traffic supply measurement, and safety controls to release off-peak capacity for other workloads.

Engineering article
2019

Taiji: Managing Global User Traffic

Co-author · SOSP ’19

A constraint-optimization system that balances data-center utilization and network latency while adapting global traffic routing to demand and failures.

SOSP ’19 paper

Selected Publications

A selection of peer-reviewed systems research. See the complete record on DBLP or Google Scholar.

  3. System Infrastructure for Mobile-Cloud Convergence

    Kiryong Ha et al. · Ph.D. thesis · 2016

  4. Towards Wearable Cognitive Assistance

    Kiryong Ha et al. · MobiSys · 2014

  5. Just-in-Time Provisioning for Cyber Foraging

    Kiryong Ha et al. · MobiSys · 2013

Patents & Talks

Selected patents

Selected talks