Datathon PolyFinances 2024

1st place at PolyFinances Datathon 2024 — we built AlexIA, a generative AI platform to support financial analysts with automation, sentiment analysis, summarization, and KPI insights.

Comment

Inspiration

At the beginning of this month, my team (Bastien Dejardin, Théo Gachet, Anais Vignaux, and I) was proud to win the annual PolyFinances Datathon, competing against over 160 talented students. We also had the incredible opportunity to present our solution to financial experts at the Caisse de dépôt et placement du Québec (CDPQ) in Montreal!

The Challenge

Create a generative AI tool to support financial analysts in their work.

Our Solution

We developed AlexIA (Advanced Learning for Economic eXploration and Interactive Analysis), a comprehensive platform designed to:

  • Automate repetitive tasks to save time.
  • Perform real-time sentiment analysis with Amazon Comprehend.
  • Summarize analyses using the Bedrock Assistant Agent and its knowledge base.
  • Provide KPI insights with public data from Statistics Canada and Yahoo Finance.
  • Offer governance insights, including leader profiles and risk assessments.

During the 48 hours of the Datathon, we created a promising prototype that could evolve into a tool to help financial analysts with real challenges. In addition, we presented a detailed one-pager and pitched our solution to the jury, competing against other incredible teams who showcased diverse, innovative approaches.

Presenting at CDPQ

We were honored to showcase AlexIA at the CDPQ headquarters in Montreal, thanks to Jean-François Bérubé. Talking to financial analysts and industry experts gave us new ideas and showed us the potential of AI in finance and other industries.

Challenges we ran into

  • Building a robust prototype in just 48 hours.
  • Integrating multiple APIs and AWS services in a short timeframe.
  • Designing a clear and intuitive interface for financial analysts.

Accomplishments that we’re proud of

  • Winning 1st place at the PolyFinances Datathon.
  • Delivering a working generative AI prototype in less than two days.
  • Presenting to financial experts at CDPQ.

What we learned

  • The power of combining generative AI with domain-specific knowledge.
  • The importance of teamwork and communication under pressure.
  • How to design AI solutions tailored for real business use cases.

Built With

Share this project:

Updates