Peter Song

Peter Song is a machine learning, data engineer, and AI writer with a focus on practical ML workflows. He shares insights on machine learning, data tools, and MLOps.

Using Large Language Models for Back-Office Automation

by

Back-office operations have long been the unglamorous backbone of business—processing invoices, handling customer inquiries, reconciling accounts, managing contracts, and countless other repetitive tasks that keep organizations running. Large Language Models (LLMs) are now revolutionizing these operations in ways that go far beyond simple automation. Unlike traditional robotic process automation (RPA) that follows rigid scripts, LLMs … Read more

What is Gaussian Process Regression?

by

Gaussian Process Regression (GPR) represents one of the most elegant and powerful approaches in machine learning, yet it remains less understood than neural networks or decision trees. At its core, GPR is a non-parametric Bayesian approach to regression that doesn’t just predict values—it provides a full probability distribution over possible functions that could fit your … Read more

Building Data Lakes with AWS Glue and S3

by

Data lakes have become the foundation of modern data architecture, enabling organizations to store vast amounts of structured and unstructured data in its native format. Amazon S3 and AWS Glue form a powerful combination for building scalable, cost-effective data lakes that can handle everything from raw logs to complex analytical workloads. This isn’t just about … Read more

Small LLM Adoption in Startups vs Big Tech

by

The landscape of artificial intelligence deployment is undergoing a fascinating divergence. While Big Tech companies continue to push the boundaries with ever-larger language models, a quiet revolution is taking place in the startup world. Small language models—those with parameters ranging from hundreds of millions to a few billion—are becoming the weapon of choice for nimble … Read more

Gemini Pro vs Gemini Ultra Differences

by

Google’s Gemini AI platform has revolutionized how we interact with artificial intelligence, offering powerful language models that compete directly with ChatGPT and other leading AI tools. However, choosing between Gemini Pro and Ultra can be confusing, especially with Google’s evolving subscription structure. This comprehensive guide breaks down the key differences between these two premium tiers … Read more

How to Integrate Small LLMs into Existing Pipelines

by

The rise of large language models has created a misconception that bigger always means better. While frontier models like GPT-4 and Claude capture headlines, small language models (typically under 7 billion parameters) offer compelling advantages for production systems: lower latency, reduced costs, enhanced privacy, and the ability to run on modest hardware. The challenge lies … Read more

Examples of LLM Hallucinations

by

Large Language Models have become ubiquitous in our digital lives, yet they harbor a troubling tendency to fabricate information with unwavering confidence. These “hallucinations” aren’t abstract theoretical concerns—they’re real occurrences that have affected legal cases, medical advice, academic research, and everyday decision-making. By examining concrete examples across different domains, we can better understand the scope, … Read more

How Often Do LLMs Hallucinate?

by

Large Language Models have transformed how we interact with artificial intelligence, powering everything from chatbots to writing assistants. But beneath their impressive capabilities lies a persistent challenge: hallucinations. These aren’t psychedelic experiences—they’re instances where AI confidently presents false information as fact. Understanding how often this happens, why it occurs, and what it means for users … Read more

Explainable AI Techniques for Black Box Models

by

In the rapidly evolving landscape of artificial intelligence, we face a fundamental paradox. The most powerful AI models—deep neural networks, ensemble methods, and complex machine learning algorithms—often operate as “black boxes,” delivering impressive results while concealing their decision-making processes. This opacity creates a critical challenge: how can we trust and responsibly deploy AI systems when … Read more

Best Practices for Securing Machine Learning Pipelines

by

Machine learning pipelines have become the backbone of modern AI applications, processing sensitive data and making critical decisions across industries. However, as these systems grow more sophisticated, they also become attractive targets for malicious actors. Securing machine learning pipelines isn’t just about protecting data—it’s about safeguarding model integrity, preventing adversarial attacks, and ensuring compliance with … Read more