I am currently helping build the intelligence layer of athlete marketing at Javelin.
"We believe that in a world where rapid content creation and short-form content is the new media, engagement converges around authenticity, discipline, and performance."
As the interest in live sports is growing more than ever, all brands from Nike to JBL are looking to partner with the best athletes. Javelin is an AI-native operating system for this newly emergent athlete economy. Our goal is to allow athletes to own their source of truth and make it easier for brands to find athletes where they are and design campaigns that are authentic to them with infinite granularity.
Tech Stack
- The brand-side web app is a NextJS app deployed on AWS Amplify. API routes are managed by AWS Lambda and API Gateway.
- The athlete-side web app is a React Native app deployed on AWS Amplify.
- The data platform is a combination of AWS Athena, S3 Parquet files, and AWS Glue.
- Embeddings for semantic search are powered by AWS Bedrock and pgvector.
- The go-based data pipeline is powered by Task definitions and SQS.
- We have different ETL jobs for different data sources, orchestrated via cron.
- The AI-powered interface (command center) uses OpenAI's structured outputs API + SSE strraming for low latency interactions.
- For better dev-ex, we have a custom CI/CD pipeline with Github Actions with commit-hooks
- We use a combination of AWS services for scalability and performance, including AWS Lambda, API Gateway, S3, DynamoDB, and SQS.