Welcome to my blog
My name is Henry Xiao, and I’ve been part of IBM since 2023.
In my role, I focus on helping financial services organizations adopt and accelerate Data and AI solutions: from modernizing data architectures to implementing AI capabilities that drive operational efficiency, strengthen risk management, and support regulatory compliance.
On this platform, I look forward to sharing insights from the evolving AI landscape, practical guidance on working with IBM’s Data and AI offerings, and learnings from real-world engagements across financial services.
Cheers!
P.S. Outside of work, I’m an enthusiastic home cook - because building good Data & AI foundations is important, but crafting a good meal is a non‑negotiable.
Posts
-
Evaluating ML/GenAI Models
When your data is messy, incomplete, biased, or misaligned with the task, it inevitably shows up as weak or unstable evaluation metrics. This guide explains major evaluation metrics across generative AI, binary classification, and regression - including real-world examples, what “good” looks like, and remediation strategies when metrics are low.
-
Fine Tuning Generative AI models with Parameter-Efficient Fine-Tuning(PEFT) with InstructLab
Back in 2024, my team and I fine-tuned a Lllama model leveraging InstrutLab and Large-scale Alignment for Chatbot (LAB) method: generating synthetic data with a teacher model, and performing Parameter-Efficient-Fine-Tuning (PEFT) to inject IBM Risk Atlas domain expertise into the opensource generative AI model.
-
Unlocking Innovation through AI, Governance, and Data, an IBM TechXchange Workshop
We hosted the first stop of our regional TechXchange workshop at the IBM Innovation Studio in Washington, DC back in August 2025. In this blog, I will share some of my takeaways as an architect and a presenter of the workshop for financial services leaders.
-
Understanding the AI Ladder Part 2
In the last post we have learned about the AI Ladder. Let’s see it in action - in a fictitious insurance company: Henry-Insurance
-
Understanding The AI Ladder Part 1: Turning Data Into Actionable Insights and Enterprise Value
The AI Ladder: Accelerate Your Journey to AI, authored by Rob Thomas, IBM Senior Vice President, offers a practical roadmap for business leaders navigating rapid technological change. In this post, I will share some of my key takeaways and draw on my experience helping Financial Services enterprises implement Data and AI solutions to illustrate what this journey looks like in practice.
-
Looking Back: Building a Retrieval-Augmented Generation (RAG) Solution in 2023
This RAG project started as a simple side experiment with my colleague Brian Paskin back in 2023. What began as a small weekend curiosity quickly revealed how Retrieval-Augmented-Generation can help mitigate some of the short comings of Large Language Models at the time
-
Driving Compliance to BCBS239 with Data Lineage
As AI adoption accelerates in the financial industry, these trends underscore the necessity for robust data governance frameworks (GDPR, BCBS 239, etc) and the tools that support compliance with these frameworks.
-
Client Workshop: Enterprise Data Quality Management
I recently had the privilege of leading a hands-on workshop with a financial services client, focusing on Data Intelligence and Data Quality. As organizations accelerate their AI initiatives and incorporate AI worklows into their operational processes, the demand for trusted, governed, high‑quality data has never been higher.
subscribe via RSS