Back

Real-Time Financial News Service - Squawk-AI Platform

1 of 9

Real-Time Financial News Service - Squawk-AI Platform

Led development of squawk-ai, building a real-time AI-powered trading news platform with WebSocket listeners, multilingual TTS, automated event filtering, and cloud deployment.

Jan 2025 - Mar 2025
Sisuro
Real-Time Financial News Service - Squawk-AI Platform

Project Overview

Led the development of an AI-powered, real-time trading news and events platform tailored for professional traders. The platform aggregates live financial news, translates it into multiple languages, and distributes it via text, audio, and social media, enabling fast and actionable insights. Managed the project end-to-end, including architecture, team coordination, technical decision-making, and cloud deployment.

Key Features

  • Real-Time Broadcasting and translation with low latency
  • Captured trading news from Live Squawk’s WebSocket, processed data, and translated it using LLM into French
  • Automated Event Calendar & AI Decision Making:
  • Fetched trading events every 60 seconds and updated real-time values.
  • Developed an AI-driven filtering system to decide whether an event should be: Announced live via Wowza Media Player. Posted on X (Twitter) using X API, based on predefined client rules
  • AI-Driven Financial Reports Automation:
  • Extracted and processed email reports into structured JSON format
  • Automated translation and formatting before distributing reports to premium users. Streaming & Text-to-Speech:
  • Developed a multilingual TTS pipeline that converted translated news into audio for live streaming. Authentication & Payments:
  • Integrated Keycloak for secure authentication and role-based access control.
  • Implemented Stripe for subscription-based monetization.

Challenges & Solutions

The main challenge was building a real-time, AI-powered trading news platform capable of aggregating, translating, and distributing high-frequency financial news with minimal latency. To address this, I implemented WebSocket listeners for live data capture and a multilingual TTS pipeline to convert translated news into audio. Integrating AI-driven filtering and decision-making required designing robust NLP pipelines and rules to automatically determine which events should be broadcast, posted on social media, or included in reports. Cloud deployment and secure access control via Keycloak ensured scalability, reliability, and compliance with client requirements.

Outcome & Impact

The platform delivers real-time, multilingual trading news and insights, enabling traders to react instantly to market events. Automated event filtering and AI-powered report generation reduced manual effort by over 70%, and the integrated AI translation eliminated the need for a specialist, lowering operational costs by approximately 60%. Accuracy improved across all reports, allowing premium users to make faster, data-driven decisions, while the Stripe subscription model supported a scalable monetization channel.

Technologies Used

Next.jsTypeScriptFast APIOpenAIAIWebSocketsText To SpeechNatural Language ProcessingLLMDockerStripeGit

Project Links

All Projects