How Segment’s Infrastructure Leader Balances Complexity and Productivity with One Key Strategy
Explore how Segment’s infrastructure leader, Jesse Adametz, drives innovation, balancing abstraction with transparency, to empower teams, streamline operations, and ensure scalability.
I got the chance to speak with Jesse Adametz, who leads the infrastructure team at Segment, a Twilio company. With a strong background in DevOps and a passion for building efficient, scalable infrastructure platforms, Jesse has been instrumental in evolving Segment's infrastructure approach from traditional SRE to a platform-oriented model.
Speaking of emphasis on balancing transparency with control - hoop.dev lets platform teams provide secure kubectl/database access without sacrificing the native CLI experience. No new tools to learn, just add SSO and audit logs on top of your existing workflows.
Company Overview
Segment, now part of Twilio, is a customer data platform that helps businesses collect, clean, and control their customer data. As Jesse puts it:
"Segment is an infrastructure-heavy product."
Dealing with high-throughput data pipelines and distributed systems, Segment handles massive amounts of customer data, making reliability and scalability crucial to their operations.
Team Structure
Jesse leads a team of 30-40 engineers spread across the US, Canada, and Bangalore, India. The team has evolved from a traditional SRE model to a platform-oriented approach:
"We've evolved all our teams inside infrastructure to behave more like platform teams."
The infrastructure team is divided into specialized focus areas that they map to their “developer journey.”:
Developer Platform
Release Engineering
Developer Enablement
Run Time Platform
Infrastructure Compute
Infrastructure Observability
Infrastructure Security
Jesse emphasizes the importance of having a dedicated security team:
"We think having a focus internally, in addition to our centralized security teams, lets us prioritize security work always, rather than in tandem with everything else."
Technology Stack
Segment's tech stack is robust and modern:
Cloud: Primarily AWS
Container Orchestration: Kubernetes (EKS), with 100+ clusters across different regions
Message Broker: Kafka (a crucial component for Segment's data pipeline)
Databases: RDS, ElastiCache
Storage: S3
Internal Developer Portal: Backstage
Jesse highlights their philosophy on technology choices:
"We use hosted services and try to buy instead of build as much as we can. If we can’t buy, we prioritize using/contributing to open source, and only when that’s not an option, we’ll build something ourselves.
They've also developed custom tools, including:
K2: A tool for managing Kubernetes cluster access and manifest templating
Internal plugins for various platforms
Delivery Pipeline
Segment's delivery pipeline is built around the concept of a "paved path," emphasizing developer autonomy while maintaining standardization:
"If you generate your application through our internal developer portal, it'll have all these things out of the box."
Key components include:
Automated engineering laptop setup
GitHub for version control
Backstage for service templating and scorecarding
BuildKite for CI
ArgoCD for GitOps-based deployments
Jesse elaborates on their use of Backstage to measure service maturity:
"We leverage different checks across a variety of “tracks” to essentially achieve scorecarding. We can check if a file exists, or introspect the file and verify if x equals y. We're doing things around code coverage for example, where we can try and evaluate different levels of maturity."
Operational Model
Segment follows a "you build it, you run it" philosophy, promoting ownership and responsibility among product teams:
"All our engineers are on call throughout Segment. You can escalate to us, but that option is there as a backup."
On-call responsibilities: Product teams are primarily responsible for their services
Product engineer involvement: High level of involvement in infrastructure knowledge
Escalation process: Infrastructure team available for escalations, but product teams are the first line of defense
Jesse emphasizes the importance of this model:
"The best people to debug an application are the people who have the most knowledge of it. And that is the application developer and their team."
Recent Success
A significant recent success has been the implementation of their service readiness framework using Backstage. This framework allows them to gamify and track service readiness across different aspects of their infrastructure, promoting best practices and standardization across teams.
Recent Challenge
Balancing abstraction and transparency in tooling has been a ongoing challenge:
"It's a balancing act. We've seen success stories both ways in the industry. The risk is if you build too much on top, it becomes hard to debug in production."
Jesse cites an example from Twilio's history where over-abstraction made debugging in production difficult, highlighting the importance of finding the right balance.
Advice for Other SRE Teams
Jesse emphasizes the importance of balancing abstraction and transparency in tooling:
"We pair tools like kubectl with our custom tool k2 for better cluster access management. We abstract where it makes sense, but we're not hiding the underlying technologies."
He advises teams to be cautious about over-abstracting, as it can make debugging difficult in production environments. Jesse also stresses the importance of collaboration and documentation:
"If something can't be maintained when a single person leaves the team, you have a severe problem. Documentation is fundamental."
Thank you!
Huge thanks for sticking around through a full year of radio silence. I know there are plenty of newsletters competing for your attention, and I dropped the ball on providing value here.
The break wasn't entirely unproductive though. After founding hoop.dev and diving deep into access management challenges with teams at Fortune 500 and scale-ups, I've collected some fascinating insights into how modern infrastructure teams operate. The recurring themes around transparency, security, and developer experience keep surfacing in every conversation.
Jesse's interview kicks off a new series focused on real, technical challenges these teams face. No fluff, no "thought leadership" - just hard-earned lessons from folks operating at scale.