Multi-Source Signal Ingestion & Parsing

Built a unified ingestion pipeline to capture trading recommendations from multiple unstructured sources including Telegram channels, PDF reports, news articles, YouTube live streams, and Twitter. Each source had a different ingestion strategy — scraping, API pulls, or third-party tools (e.g., PDF parsing via Pulse AI). Despite differences in input format, all data was normalized into a common structure using LLM-based parsing (ChatGPT API). The system handles noisy, inconsistent text inputs and converts them into structured trade signals ready for downstream processing.