Schedule: Thursday, June 14

When Julia Grace joined Slack two-and-a-half years ago, the company had fewer than 100 engineers. It's now at more than 350, and her own team grew from 10 to 50 people in 18 months. Expanding on her keynote, Julia shares tips and stories from the leadership front lines as she transitioned from line manager to managing managers.

“That would never work here.” You’ve likely heard this sentiment (or maybe you’ve even said it yourself). Good news: change is possible. Donovan Brown explains how Microsoft's Visual Studio Team Services (VSTS) went from a three-year waterfall delivery cycle to three-week iterations and open sourced the VSTS task library and the Git Virtual File System.

Engineering teams want technically competent managers, but they also often want managers to keep their hands off their code. So how can managers keep their technical skills relevant in order to add the most value? Kathleen Vignos shares creative strategies for developing and maintaining technical skills—some through the act of managing itself.

Sunil Sadasivan compares the work environments of startups to those of bureaucracies and shares lessons for maintaining an optimal engineering work culture learned at the US Department of Veterans Affairs.

Homogenous teams are one proven cause of missteps and flaws in software products and pipelines. Sacha Judd offers a fresh perspective, detailing available tools to improve hiring, promotion, and internal culture.

In 2007, Pat Helland published "Life Beyond Distributed Transactions: An Apostate’s Opinion," in which he conducts a thought experiment on how to design a distributed database that can scale almost infinitely. While the paper explicitly addresses distributed database design, Sean Allen shows that the ideas are far more widely applicable, particularly in scaling stateful applications.

Manish Mehta and Torin Sandall lead a deep dive into how Netflix enforces authorization policies (“who can do what”) at scale in its microservices ecosystem in a public cloud without introducing unreasonable latency in the request path.

Performance debugging is a crucial part of ensuring code is production ready, particularly as a company and its products grow. However, bottlenecks that hold these services back can be hard to identify. Christian Grabowski shares his experience debugging bottlenecks in distributed systems, at both a macro (metrics, distributed tracing) and a micro (user space and kernel space profiling) level.

Kyle Kingsbury offers an overview of Tesser, a Clojure library for writing commutative, parallel folds that can be chained and composed into complex single-pass reductions that are dramatically faster on multicore systems and can be transparently distributed over Hadoop.

Tyler McMullen offers an overview of sandboxing compilers, which provide important benefits but are also challenging to make both safe and fast. Tyler covers machine code generation and optimization, trap handling, and memory sandboxing and illustrates how to integrate them into an existing system—all based on a real compiler and sandbox, currently in development.

Local service discovery and availability is easy, but how do you discover services in other data centers or other cloud providers? Seth Vargo explains how HashiCorp Consul can provide service discovery, monitoring, and failover across many regions and multiple public and private cloud providers.

Ryan Neal explains how Netlify planned, tested, and executed its first multicloud migration that could direct traffic to Google Cloud (GCP), AWS, and Rackspace Cloud on demand, without any service interruptions. Along the way, Ryan shares lessons learned and key takeaways you can apply to your own infrastructure decisions.

Jessica DeVita tells the story of how a team at Microsoft challenged themselves to retrospect their retrospectives and shares what they learned about applying human factors ideas to software development.

Machine learning (ML) is becoming a critical skill for developers and businesses alike. Applying ML successfully in real-world scenarios, however, remains challenging. Alex Jaimes discusses how to find opportunities to apply ML, the pitfalls in applying it, and the steps required to succeed—from data to metrics to testing to other critical factors.

Paul McCallick discusses how and why Nordstrom has moved to an only-production viewpoint, saving countless engineering cycles and putting effort where it matters.

There are many conference sessions on "how to get started with X." But once you've gotten up and running, there isn't always a lot of guidance on how to solve harder issues. Abby Fuller takes you beyond getting started with containers on AWS, covering advanced topics like hybrid clusters, bringing your own AMI, working with Docker settings not supported in the UI, and debugging load balancers.

Liz Rice leads a dive into what's easy—and what's not—about finding and patching security vulnerabilities in containers.

Qingyang Chen and Appu Goundan demonstrate how to speed up container-based development by building container images with Jib, a Google image build tool for Java applications.

David Cheney shares real-world advice on how to extend the capabilities of a Kubernetes cluster, using the development of the open source Contour Ingress controller as a case study.

Modern microservices architectures (like those run on Kubernetes) need modern security solutions to provide least-privilege security. Cynthia Thomas outlines traditional firewall methods and details the evolution of the distributed security model to enforce least privilege for microservices.

Mike Roberts leads a warts-and-all journey through some of the limitations of a serverless approach and shares a practical set of techniques for dealing with these concerns.

Serverless data access (via SQL and other data query/processing languages such as Spark) is fast becoming the norm. Lynn Langit compares the state of public cloud serverless SQL via AWS Athena, Google Big Query, and others and explores architectural patterns and examples of services for emerging serverless and data lake cloud pipelines.

Tooling is necessary for serverless and service-full applications. Donna Malayeri shares a decision framework for choosing infrastructure deployment tools, based on whether you need flexibility and control or simplicity and ease-of-use. You'll learn how to evaluate several popular cloud automation tools, including AWS SAM, Terraform, Chalice, Serverless Framework, and more.

FaaS functions are great for small functionality but not for complex real-world applications. Soam Vasani and Timirah James explore four available options for composing functions, along with a deep dive into workflows.

Serverless and other stateless applications still manipulate state—somewhere. Erica Windisch explains why observing this state and knowing where, how, and why that state is manipulated is important for operational security and developer concerns such as debugging.

Mark Prichard reviews available metrics from infrastructure, Kubernetes, containers, and application code and shares options for viewing them holistically, thus providing a complete picture of how your applications are behaving and how users are experiencing them.

Much of the monitoring data we rely on is fundamentally flawed, lacking the resolution and accuracy needed to effectively detect and diagnose many issues. Digital signal processing science has overcome similar challenges for audio. Using sound as an example, Jon Hodgson explains how these principles are leveraged by organizations to improve the fidelity of their performance monitoring.

Program chairs Nikki McDonald, Ines Sombra, and James Turnbull open the second day of keynotes.

Kyle Kingsbury explores anomalies in three distributed systems—Tendermint, Hazelcast, and Aerospike—and shares general strategies for correctness testing using Jepsen, a distributed system testing harness that applies property-based testing to databases to verify their correctness claims during common failure modes: network partitions, process crashes, and clock skew.

Kyle York and Richard Lee explore Netra’s high-performance computing environment, focusing on how the company's AI and deep learning models process tens of millions of images and videos each day in a time- and cost-effective manner. Along the way, they explain what worked, what didn't, and why you need an Agile, hybrid infrastructure if you want to build an AI business at the scale of social.

Declarative application management enables developers and operators to simplify their configurations while deploying into increasingly complex environments. Bryan Liles explains how to evaluate and integrate these new practices into existing continuous integration pipelines.

Change is inevitable, but the aftereffects can be both good and bad. Having the right tools is one way to meet this challenge. Dave Andrews explains how to wield the power of a global 50 Tbps application delivery network, featuring 125+ points of presence, to ensure maximum availability during and after a change.

Nicole Forsgren shares results and stories from four years of research to uncover the secrets and surprises of what really makes high-performing technology-driven teams and organizations.

Artificial intelligence has been almost here for 50 years, but we don't need to wait for it to escape the laboratory. Adding a manageable dose of actionable intelligence to your operations management workflow can save you time and aggravation. PagerDuty will talk about how AI's limitations and how it can decrease your noise and suggest possible courses of action.

When Julia Grace joined Slack two-and-a-half years ago, the company had fewer than 100 engineers. It's now at more than 350, and her own team grew from 10 to 50 people in 18 months. Julia shares tips and stories from the leadership front lines as she learned how to rapidly scale herself and her leadership team during a period when her job was substantially changing every six months.

Jumpstart your networking at Velocity at Speed Networking before the keynotes begin. Bring your business cards and prepare a minute of patter about yourself, your projects, and your interests.

Join other attendees during lunch at Velocity to share ideas, talk about the issues of the day, and maybe solve a few. Not sure which topic to pick? Don’t worry—it's not a long-term commitment. Try two or three and settle on a different topic tomorrow.

Join us for the closing celebration of Velocity and Fluent. Don’t miss this last chance to mingle.