DraftKings & MongoDB: Supercharging Engineering with AI

Overview

This presentation explores how DraftKings, a leading digital sports entertainment company, is leveraging MongoDB and AI/ML technologies to dramatically increase engineering speed, empower internal builders, and evolve their architecture. Key topics include:

• DraftKings' journey from a monolithic application to a microservices-based, data-focused architecture
• Adoption of MongoDB to enable flexible data models and real-time inference
• Embracing no-code/low-code tools and AI-powered automation to boost productivity
• Upskilling engineering teams to effectively leverage AI capabilities
• Balancing speed and reliability in mission-critical sports betting workloads

DraftKings' Technical Transformation

From Monolith to Microservices

• DraftKings started with a monolithic application focused on fantasy sports
• Struggled to scale the monolith to handle large traffic spikes (e.g. 10x increase in 10 seconds after a touchdown)
• Transitioned to a microservices architecture, then migrated to Kubernetes for better scalability
• Moved from a Microsoft-centric stack to a more data-focused approach

Adopting MongoDB

• Acquired a company that was a heavy MongoDB user, which exposed DraftKings to the benefits of the document-oriented database
• Enabled engineering teams to use the "right tool for the job" rather than a single, mandated database
• Leveraged MongoDB's flexible data model and real-time capabilities to support sports betting workloads

Evolving the Data Architecture

• Shifted from a traditional data warehouse approach for batch-based ML to real-time inference in customer-facing flows
• Utilized MongoDB's vector search capabilities to enable content search and personalization in a CMS application
• Found MongoDB to be a better fit than SQL-based databases for the flexibility and speed required in sports betting workloads

Embracing No-Code/Low-Code and AI Automation

Empowering Internal Builders

• Recognized the potential of no-code/low-code tools to enable a broader set of "builders" beyond the engineering team
• Implemented tools like N8 to allow non-technical employees to automate workflows and experiment with AI-powered solutions
• Focused on enabling engineers to use AI coding tools and understand their capabilities through hands-on experimentation

Balancing Speed and Reliability

• Prioritized uptime and resilience as the most critical feature for DraftKings' customer-facing applications
• Carefully evaluated the impact of no-code/low-code solutions on production systems, using metrics like human toil reduction and usage metrics as proxies for value
• Defined a "graduation path" to move successful no-code prototypes into more robust, production-ready solutions

AI-Powered Automation

• Developed an AI-powered code review agent using Anthropic and AWS Bedrock
• Saw significant improvements in code review quality and speed, with junior developers benefiting from immediate feedback
• Observed more senior engineers embracing AI tools to increase their own productivity, rather than seeing it as a threat to their skills

Lessons and Advice

Upskilling Engineering Teams

• Focused on enabling engineers to use AI tools themselves and understand their capabilities through hands-on experimentation
• Leveraged a hybrid approach, with data science/ML experts educating other teams and engineers sharing knowledge within their own teams
• Recognized that the most senior engineers were often the quickest to adopt AI tools, as they could better leverage the technology to increase their own productivity

Overcoming Legacy Challenges

• Advised working closely with business stakeholders to understand the real need behind legacy processes and find ways to co-invest in modernization
• Emphasized the importance of "paying down the debt" and making time for innovation, rather than just trying to bolt on new technologies
• Recommended meeting developers where the toil is and focusing on automating those workflows first

The Road Ahead

• Continues to explore ways to leverage AI and MongoDB to drive faster innovation and better customer experiences, without compromising on reliability
• Plans to expand into new areas like prediction markets, using AI to enable more personalized and responsive customer interactions
• Remains committed to enabling the right tool for the job, whether that be MongoDB, Aurora, or other technologies, while maintaining a strong partnership with AWS and MongoDB

Key Takeaways

• DraftKings' journey demonstrates the power of embracing flexible, data-centric architectures and empowering a broader set of "builders" through no-code/low-code tools and AI automation
• Careful evaluation of the impact and risk of new technologies, with a focus on uptime and reliability, is crucial for mission-critical applications like sports betting
• Upskilling existing engineering teams to effectively leverage AI capabilities is a key challenge, but can be addressed through hands-on experimentation and a hybrid approach
• Overcoming legacy technical debt requires close collaboration with business stakeholders and a willingness to "pay down the debt" to create space for innovation
• DraftKings' continued partnership with AWS and MongoDB highlights the value of strategic technology alliances in driving digital transformation