Contents# Introduction# Setting a Complexity Budget with Textstat# Implementing the LangChain Pipeline# Wrapping Up # Introduction  Large language models (LLMs) have a taste for using “flowery”, sometimes overly verbose language in their responses. Ask a simple question, and chances are you may get flooded with paragraphs of overly detailed, enthusiastic, and complex prose. This usual behavior is rooted in …

Guardrails for LLMs: Measuring AI ‘Hallucination’ and Verbosity Read More »

any time in the data engineering world, you’ve likely encountered this debate at least once. Maybe twice. Ok, probably a dozen times😉 “Should we process our data in batches or in real-time?” And if you’re anything like me, you’ve noticed that the answer usually starts with: “Well, it depends…” Which is true. It does depend. But “it …

Batch or Stream? The Eternal Data Processing Dilemma Read More »

  Contents# Introduction# Why JSON Wastes Tokens in LLM Pipelines# What TOON Actually Is and When It Is Worth Using# Getting Started with TOON# Final Thoughts # Introduction  JSON is great for APIs, storage, and application logic. But inside large language model (LLM) pipelines, it often carries a lot of token overhead that does not add much value to the model: …

Stop Wasting Tokens: A Smarter Alternative to JSON for LLM Pipelines Read More »

(not that long ago) when being a data scientist meant living in a notebook, tweaking hyperparameters as if your life depended on it, and in a lot of cases, the whole project did, indeed, depend on it. Do you remember those overnight grid searches? Or building feature engineering pipelines that felt more like art than …

From Data Scientist to AI Architect Read More »

  Contents# Introduction# Initial Setup# Checking Univariate Normality# Checking Multivariate Normality# Checking Homoscedasticity# Checking Sphericity# Checking Multicollinearity# Wrapping Up # Introduction  Anyone who has spent a fair amount of time doing data science may sooner or later learn something: the golden rule of downstream machine learning modeling, known as garbage in, garbage out (GIGO). For example, feeding a linear regression model with highly collinear …

Building Modern EDA Pipelines with Pingouin Read More »

Across 64 English authorities and six 2026 scenarios, even the strongest scenario shock was only 13% of the median uncertainty band. In plain English: the model’s assumptions moved the result less than historical forecast error did. The most aggressive challenger surge I could parameterise sits inside the noise the model has produced in past elections. …

When the Uncertainty Is Bigger Than the Shock: Scenario Modelling for English Local Elections Read More »

  Contents# Introduction# Bootleggers and Baptists# Why the Coalition Works So Well# Learning to Tell Them Apart# Final Thoughts # Introduction  Every organization loves to call itself “data-driven.” It’s become the gold standard of credibility, the thing you say to shut down dissent in a meeting. But here’s something worth sitting with for a second: the phrase “according to data analytics” …

Baptists and Bootleggers: The Hidden Coalition Behind ‘Data-Driven’ Decisions Read More »

ContentsAI AgentsComponents of an AI AgentReAct (Reasoning + Acting) in AgentsA basic ReAct workflow in an AI agent usually looks like this:Structure of AI AgentsSingle Agent vs Multi-AgentWhen to Build A Multi-Agent SystemWalkthrough of A Multi-Agent ProjectMulti-Agent ArchitectureMemoryOrchestratorProject SetupDemo Video of the Multi Agent Agent RAG ResearcherNotesConclusionReach to me via:References AI Agents When building an …

Single Agent vs Multi-Agent: When to Build a Multi-Agent System Read More »

Image by Editor   Contents# Introduction# What Is Voxtral TTS?# Cloning a Voice from Three Seconds of Audio# Latency Performance: Built for Real-Time Conversations# How Voxtral TTS Works# Getting Started: Installation and Setup# Voice Cloning with a Custom Voice: A Practical Example# Use Cases# Licensing and Deployment Considerations# Conclusion # Introduction  Voice-enabled applications are everywhere, from virtual assistants to customer service chatbots. But for developers, building …

Open Weight Text-to-Speach with Voxtral TTS Read More »

Authors: Ahsaas Bajaj and Benjamin S Knight ? We ran 134,400 simulations grounded in real production ML models to find out. The answer depends on what you’re optimizing for, and on a single diagnostic you can compute before fitting a model. If you’ve ever trained a linear model in scikit-learn, you’ve faced this question: RidgeCV, …

Which Regularizer Should You Actually Use? Lessons from 134,400 Simulations Read More »