X.com stock news analysis

Built a time-boxed prototype that ingests stock and company-related posts from X.com, stores the content with provenance metadata in S3, and enables users to ask questions in Amazon Q Business via a RAG workflow with source citations.

Role

Principal engineer

Timeframe

June 2025 (time-boxed engagement)

Outcomes

• Delivered a natural-language-queryable proof of concept with source citations.
• Funded by Amazon and delivered as a working baseline for future expansion.

Stack

• AWS Lambda, S3, SQS, DynamoDB
• AWS Bedrock, Amazon Q Business
• Custom Q Business connector

Writeup

Implemented X.com ingestion from a curated set of trader profiles, extracted cashtags and sentiment, and persisted content with provenance metadata for attribution. Built an S3 to SQS to Lambda pipeline, added leaky-bucket rate limiting, and integrated a custom Amazon Q Business connector to keep the RAG index in sync with the corpus.