Towards Data Science

Leveraging the power of AI tools comes with a range of complications when applied to IoT systems. Illia Smoliienko unpcaks the challenges and proposes strategies for overcoming them. towardsdatascience.com/how-ai-tools-g…

A new paper walkthrough by Muhammad Ardi is always a cause for celebration — don't miss his latest, which unpacks the inner workings of the cross-state partial network paper, and includes a full PyTorch implementation. towardsdatascience.com/cspnet-paper-w…

"While a model pauses to think, it generates hidden reasoning tokens. These tokens never appear in the final chat bubble, but they represent a massive surge in billable compute on your monthly invoice." Mostafa Ibrahim explains why teams should be mindful when scaling the use of reasoning models, and presents a framework that will help you avoid unpleasant surprises. towardsdatascience.com/inference-scal…

We're thrilled to welcome Ahsaas Bajaj and Benjamin S Knight to TDS! Don't miss their excellent debut article on choosing the right regularizer, based on wide-scale simulation experiments they conducted. towardsdatascience.com/which-regulari…

TurboQuant has been generating a lot of buzz in recent months. Amit Portnoy points to an earlier quantization method (of which he's a co-author) and examines how the latter still outperforms the more recent arrival. towardsdatascience.com/how-a-2021-qua…

I got it wrong. A bug flipped my entire result: fragmentation didn’t rise in 66/67 councils. It rose in just 18. What actually changed? Voter churn doubled. The party system didn’t fragment. @TDataScience towardsdatascience.com/fractured-loca… @tableau public.tableau.com/app/profile/ob…

Create, migrate, tune, run PoCs on unlimited PostgreSQL DBs in the cloud, using your favourite AI coding agent. That's what @ghostdotbuild promises. To find out more, read my latest article on @TDataScience, where I deep dive into this useful product . towardsdatascience.com/ghost-a-databa…

Under what circumstances might strong metrics not necessarily mean that a model is ready for deployment in a real-world environment? @erikapatg shares a sharp and actionable reflection on the importance of robust methodology. towardsdatascience.com/why-powerful-m…

"Ghost is especially useful for testing, prototyping, agent workflows, branch databases, migration experiments, and disposable database environments." @taupirho presents a hands-on introduction to Ghost, an agent-friendly database. towardsdatascience.com/ghost-a-databa…

For his latest deep dive, @DataSenseiObi presents a data-quality case study on English local elections, covering categorical normalisation, metric validation, and why raw labels should never define analytical groups. towardsdatascience.com/fractured-loca…

"If you’re applying for junior roles right now, you’ve probably noticed something weird: the listings still exist, but the doors just feel heavier to push." Ivo Bernardo shares actionable insights on getting hired in the AI era. towardsdatascience.com/how-to-get-hir…

The real value of agents comes from how broadly you apply them. @EivindKjos breaks down how Claude Code can extend beyond programming into general productivity. towardsdatascience.com/how-to-apply-c…

Most AI systems fail not at retrieval, but at what gets passed into the context window. Emmimal P Alexander explains why a dedicated context engine is essential for reliable LLM systems. towardsdatascience.com/rag-isnt-enoug…

Most users barely scratch the surface of Claude Cowork’s capabilities. @EivindKjos explores how to fully integrate it into your engineering and productivity processes. towardsdatascience.com/how-to-maximiz…

"Abstractions trade that visibility for speed. That’s a fair trade at first, until the hidden complexity becomes the very thing you need to understand." @BenjaminNweke11 explains why AI practitioners are moving towards building agent orchestration layers themselves, instead of relying on readymade frameworks. towardsdatascience.com/why-ai-enginee…

Why do monotonicity and stability matter for scoring models? @jumbong_junior presents a hands-on tutorial on using Python to validate variables' consistency. towardsdatascience.com/how-to-study-t…

What happens when a RAG pipeline "looks at a document as a hierarchical tree of semantic blocks, rather than a bag-of-words that needs to be shredded blindly into chunks to answer a query"? Partha Sarkar explores the potential of multimodal RAG in his latest deep dive. towardsdatascience.com/proxy-pointer-…

"We’ll see why the obvious approach doesn’t work, walk through the four standard ways to handle uncertainty in a linear program, and finish with a quick sanity check on whether any of this is worth the effort." Berend Markhorst unpacks the basics of stochastic programming. towardsdatascience.com/a-gentle-intro…

Beyond automation, collaboration is where Claude Cowork becomes interesting. @EivindKjos shows how to leverage it effectively alongside Claude Code. towardsdatascience.com/how-to-maximiz…

Claude Code is not limited to writing code. @EivindKjos shows how it can be applied to a wide range of non-technical tasks across your workflow. towardsdatascience.com/how-to-apply-c…