December 2024

Building AI-Driven Trading Systems at Intact Financial

How I developed scalable data pipelines and integrated LLMs for cryptocurrency trading, achieving significant performance improvements.

The Challenge

When I joined Intact Financial Corporation as an AI Engineer in 2023, the cryptocurrency trading data processing was becoming a bottleneck. The existing system couldn't handle the volume and velocity of real-time market data effectively.

The Solution

I designed and built scalable data pipelines that revolutionized how we process cryptocurrency trading data:

  • 65% Performance Improvement: Through automation and optimization of data processing workflows
  • LLM Integration: Implemented GPT, Claude, and Hugging Face models for trading signal generation
  • Multimodal AI: Developed systems for market analysis, financial report NLP, and trading visualization
  • Voice Synthesis: Created audio-based trading alerts and market summaries

Technical Implementation

The architecture leveraged:

  • Cloud Infrastructure: AWS and GCP for scalable compute and storage
  • MLOps Pipelines: Terraform and Kubernetes for deployment and monitoring
  • Data Processing: Real-time streaming with automated drift detection
  • API Development: Modular backend APIs for trading systems integration

Impact & Results

The new system not only improved performance but also enabled:

  • Real-time market analysis with LLM-powered insights
  • Automated trading signal generation based on multiple data sources
  • Robust monitoring and alerting systems
  • Seamless integration with existing trading infrastructure

Key Learnings

This project taught me the importance of:

  • Designing for scale from day one
  • The power of LLMs in financial applications
  • Cross-functional collaboration in high-stakes environments
  • Building production-grade AI systems with proper monitoring
Previous Post
← Blog Overview
Next Post
Blog Overview →