1. Reading Financial Documents with AI
    • Banks run on documents: quarterly reports, compliance certificates, credit memos, broker research, and borrower spreadsheets. Multimodal LLMs can now read all of them, but reading is not the hard part. The hard part is knowing when a number is right. This post explains why financial documents are hard for machines, the difference between retrieving and extracting, three architecture archetypes for AI-based document ingestion (with Python code), how to measure extraction quality, and how to choose the right approach for your use case.

  2. Evaluation Metrics for Real-Time Financial Fraud Detection ML Models
    • After training a real-time financial fraud detection model, the next step is to evaluate its performance. This post provides an overview of the most common evaluation metrics and considerations for fraud detection models, including confusion matrix, precision, recall, F1-score, AUC-ROC, AUC-PR, and more.

  3. From Classical ML to DNNs and GNNs for Real-Time Financial Fraud Detection
    • Financial transaction fraud is a pervasive problem costing institutions and customers billions annually. This survey reviews the current state-of-the-art in real-time transaction fraud detection, spanning both academic research and industry adopted solutions.

  4. Building GenAI Applications Today
    • Generative AI has taken the world by storm, offering endless opportunities for innovation. But as with any new technology, there are plenty of pitfalls to avoid. In this post, I dive into the current state of AI startups, shedding light on what works, what doesn't, and why. If you're looking to build something with AI, let’s step back and reflect on where it makes sense to innovate and where it doesn't. From avoiding overhyped general-purpose chatbots to understanding the limitations of AI in niche markets, this post offers practical insights to help you navigate the AI hype more effectively.

  5. Why GenAI Will NOT Replace Software Engineers
    • GenAI systems are becoming more and more capable of performing complex cognitive tasks that were once thought to be uniquely human. In particular, LLMs are proven to be very good at writing code. With all the buzz around GenAI replacing software engineers, are our jobs really at risk? In this article, I dive deep into AI's current capabilities and how they stack up against the demands of real-world software development. I discuss the actual potential of GenAI to assist in common development activities, and where it still falls short when it comes to creative problem solving and human intuition.

  6. AI Doesn't Make Me the Same Coffee
    • Advancements in AI are set to transform the way we interact with the world, offering increased free time and enhanced capabilities. However, if not carefully managed, AI poses risks such as reducing human experiences and potentially becoming a significant social issue.

  7. Surviving the AI Revolution as a Software Engineer
    • The next wave of extremely powerful and specialized AI models is coming, and it will likely revolutionize software engineering as we know it. As developers, we need to get the skills to surf the wave and keep our jobs. As with every technological breakthrough, AI will create new jobs and destroy others. In this article, I will focus on how you can keep your job in this new world dominated by AI. My focus is on software engineering jobs, but most of the principles I mentioned also apply to other professions.