CarGurus’ journey building a developer portal and increasing AI adoption

Key takeaways

CarGurus successfully transitioned from a monolithic architecture to microservices using an IDP framework that streamlined operations.
The five pillars of the Showroom IDP—discoverability, governance, self-service ability, transparency, and operational efficiency—were critical in enabling developers to work better.
AI coding assistants are being evaluated for their effectiveness, and qualitative feedback from developers will guide the decision-making process.
Engagement from leadership is crucial for driving the adoption of new tools like AI coding assistants.
Measuring success through various metrics such as flow, efficiency, job satisfaction, and time savings ensures that adoption translates into tangible benefits.

Summary

CarGurus’ Internal Developer Portal Journey

Frank Fodera discusses the inception of the internal developer portal (IDP) at CarGurus, named Showroom, which began in 2019 as a response to ownership issues while transitioning to a microservices architecture. He explains how the initial service catalog, originally managed through spreadsheets and wikis, evolved into a structured IDP with components like compliance, self-service capabilities, and operational efficiency. The goal was to centralize information for developers, streamlining the way they access resources and ownership of services.

The Five Pillars of Showroom

Frank outlines the five major pillars of the Showroom IDP: discoverability, governance, self-service ability, transparency, and operational efficiency. Each pillar addresses specific aspects of developer experience, ensuring that information is centralized, compliance rules are adhered to, and developers can perform actions efficiently. This structure aims to enhance productivity and reduce cognitive load for developers.

Shifting from Monolith to Microservices

The conversation delves into CarGurus’ challenging transition from a monolithic architecture to microservices, emphasizing early struggles and a shift to a Strangler Fig pattern for gradual decomposition. The IDP played a pivotal role in this transition by providing a supportive framework that facilitated faster service development, reducing deployment times from 75 days to under three days.

AI Coding Assistant Adoption

As part of its commitment to efficiency, CarGurus is exploring various AI coding assistants with ongoing evaluations. Frank shares insights into how the company is conducting a bake-off between three tools to determine the best fit for their needs. He emphasizes the importance of qualitative feedback from developers and the necessity to measure both the efficiency gains and any potential drawbacks, such as burnout.

Leadership and Adoption Strategies

The podcast wraps up with a discussion on strategies for driving AI adoption within the organization. Frank highlights the importance of leadership involvement in encouraging tool usage and setting clear goals for efficiency improvements. The use of metrics, education, and leveraging champions within the organization helps promote adoption and ensures that developers are supported in their transitions to using these new tools.

Show notes

In this episode, Abi Noda talks with Frank Fodera, Director of Engineering for Developer Experience at CarGurus. Frank shares the story behind CarGurus’ transition from a monolithic architecture to microservices, and how that journey led to the creation of their internal developer portal, Showroom. He outlines the five pillars of the IDP, how it integrates with infrastructure, and why they chose to build rather than buy. The conversation also explores how CarGurus is approaching AI tool adoption across the engineering team, from experiments and metrics to culture change and leadership buy-in.

Where to find Frank Fodera :

• LinkedIn: https://www.linkedin.com/in/frankfodera/

Where to find Abi Noda:

• LinkedIn: https://www.linkedin.com/in/abinoda

In this episode, we cover:

(00:00) Intro: IDPs (Internal Developer Portals) and AI

(02:07) The IDP journey at CarGurus

(05:53) A breakdown of the people responsible for building the IDP

(07:05) The five pillars of the Showroom IDP

(09:12) How DevX worked with infrastructure

(11:13) The business impact of Showroom

(13:57) The transition from monolith to microservices and struggles along the way

(15:54) The benefits of building a custom IDP

(19:10) How CarGurus drives AI coding tool adoption

(28:48) Getting started with an AI initiative

(31:50) Metrics to track

(34:06) Tips for driving AI adoption

Referenced: