What tech stack does Netflix use?

Netflix's backend is built primarily in Java, using Spring Boot as the service framework since a broad internal migration approximately 2016–2018. Python is heavily used for data science, machine learning, and internal tooling. The company operates thousands of microservices communicating via GraphQL — using Netflix's open-source DGS (Domain Graph Service) framework built on top of Spring Boot — and RESTful APIs. Node.js appears in secondary services and tooling based on job postings.

Netflix completed its migration to AWS in 2016 and now runs across four active global AWS regions, using EC2 for compute, S3 for object storage, CloudFront for edge delivery, EKS for container orchestration, EMR for managed Hadoop/Spark, DynamoDB for low-latency key-value data, and Redshift for analytical queries. Alongside AWS-managed services, Netflix built and open-sourced Titus (container management, now a CNCF project), Spinnaker (multi-cloud continuous delivery, now managed by the CD Foundation), and Zuul (API gateway). Open Connect — Netflix's proprietary CDN using ISP-embedded appliances deployed in approximately 1,000+ locations worldwide — handles last-mile video delivery, reducing AWS bandwidth costs and improving stream quality at scale; this is one of Netflix's most strategically significant proprietary infrastructure investments.

Netflix's web frontend is built in React, with the DGS GraphQL framework providing the API contract between frontend and backend microservices. Mobile apps are native: Swift for iOS and Kotlin for Android, following modern native-first patterns that prioritize performance for video playback.

Netflix's data platform is one of the most sophisticated in the industry. Apache Kafka handles real-time event streaming at massive scale (trillions of events per day); Apache Flink processes streaming analytics; Apache Spark powers batch processing and ML model training; Apache Iceberg provides the open table format for Netflix's data lake (enabling schema evolution and time-travel queries at petabyte scale); Redshift and Apache Druid serve analytical queries; and Presto/Trino enables ad-hoc SQL exploration at petabyte scale. The Apache Iceberg project itself was created at Netflix and later donated to the Apache Software Foundation. Tableau is used for business analytics visualization. For business tooling, Netflix uses Atlassian (JIRA and Confluence) for engineering project management and Google Workspace for corporate productivity. OneLogin handles SSO and identity management across the organization.

Netflix's build-first engineering culture means it has historically built custom replacements for categories where commercial tools did not meet its scale or flexibility requirements — monitoring (Atlas), CDN (Open Connect), container orchestration (Titus), continuous delivery (Spinnaker), workflow orchestration (Conductor, now open-source), and the Iceberg table format. Vendors must demonstrate clear differentiation over what Netflix could build internally, and engineers will conduct technically rigorous proof-of-concept evaluations before recommending any external purchase.

Categories where Netflix is most likely to buy commercial products include: security and compliance tooling (SOC 2, identity, vulnerability management — Netflix cannot cost-effectively build everything in security), advertising technology (its own ad stack is maturing but still acquiring SSP and measurement partners), data visualization (Tableau confirmed), communication and collaboration (Google Workspace), and HR/legal/finance platforms that are category-standard. Vendors that are open-source friendly, provide strong AWS-native integrations, and can demonstrate performance at Netflix's scale of tens of millions of concurrent streams will resonate most with the engineering evaluation team.

The displacement opportunity for core infrastructure (compute, CDN, container orchestration) is essentially zero — Netflix has already built custom solutions. The best entry points are point solutions in the security, observability, and developer experience categories that complement rather than replace the Netflix OSS foundation. Advertising technology vendors should closely monitor Netflix's internal ad tech build-out: the company's stated goal of a first-party programmatic stack may displace existing SSP and measurement partners by 2027–2028.