Lightnews — Scholar-powered news

Towards Data Science @towardsdatascience.com · 12h

See why a well-intentioned optimization can sometimes hurt performance. Ryan Pégoud's new article shares a surprising benchmark result from a Triton GEMM kernel and discusses the importance of profiling.

Learning Triton One Kernel at a Time: Matrix Multiplication | Towards Data Science

Tiled GEMM, GPU memory, coalescing, and much more!

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 13h

Learn to translate complex model metrics into clear business KPIs. This new article from Kristopher McGlinchey teaches how to frame your results in terms of money saved, generated, or time saved.

Building A Successful Relationship With Stakeholders | Towards Data Science

Show your value by moving beyond the technical

towardsdatascience.com

1

Reposted by Towards Data Science

dmitryontech.bsky.social @dmitryontech.bsky.social · 2d

How to Build Effective Agentic Systems with LangGraph in @towardsdatascience.com

towardsdatascience.com/how-to-build...

How to Build Effective Agentic Systems with LangGraph | Towards Data Science

Create AI workflows with agentic frameworks

towardsdatascience.com

1

Reposted by Towards Data Science

taupirho.bsky.social @taupirho.bsky.social · 1d

In my recent article on @towardsdatascience.com , I dive into how you can call Mojo directly from Python to supercharge your code without leaving the Python comfort zone. Read it for FREE using this link.
towardsdatascience.com/python-can-n...

Python Can Now Call Mojo | Towards Data Science

Boost your runtimes with lightning-fast Mojo code

towardsdatascience.com

1 1

Reposted by Towards Data Science

roadmap.sh @roadmapsh.bsky.social · 18h

Want to stand out as a junior developer? 🙋 Knowing #SQL is non-negotiable. With our SQL course, you’ll...

✔️ Go from SELECT to complex joins
✔️ Practice in an integrated IDE
✔️ Earn a certificate to share with employers

Start learning → roadmap.sh/courses/sql

1 2

Towards Data Science @towardsdatascience.com · 15h

🌟 Welcome Nidhin Karunakaran Ponon! 🌟 Nidhin dives into practical guardrails for AI systems — what they are, why they matter, and how to create them in your own workflows.

Want to see your work featured next? Submit your article and join our community of paid TDS contributors: bit.ly/TDSContributor

How To Build Effective Technical Guardrails for AI Applications | Towards Data Science

Exploring the most practical guardrails to implement at ground level

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 16h

Struggling with unreliable AI predictions for equipment failure? It's often due to a lack of specific training data. Illia Smoliienko's debut TDS article explains why this makes human analysts essential in predictive maintenance.

Why AI Still Can’t Replace Analysts: A Predictive Maintenance Example | Towards Data Science

Learn about the limitations of AI in analytics through the example of bearing vibration data analysis

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 18h

Discover why a programmer's core function becomes more valuable, not less, in the age of AI. This new article by Elisha Rosensweig and Eitan Wagner explores how the process of coding is one of discovery, making crucial decisions that AI's statistical approach cannot replicate.

Human Won’t Replace Python | Towards Data Science

Why vibe-coding is not a step up from "classic" coding — and why it matters

towardsdatascience.com

1

Towards Data Science @towardsdatascience.com · 18h

Reduce development friction and standardize project setup across your team. Elena Jolkver's new article walks you through Cookiecutter, a tool for creating project templates. Learn how to define a reusable structure, automate boilerplate, and generate fully-formed projects with a single command.

How to Spin Up a Project Structure with Cookiecutter | Towards Data Science

If you’re anything like me, “procrastination” might as well be your middle name. There’s always that nagging hesitation before starting a new project. Just thinking about setting up the project…

towardsdatascience.com

1

Towards Data Science @towardsdatascience.com · 19h

Eivind Kjosbakken explains how to leverage LangGraph to build effective agentic systems, simplifying AI workflows while managing tool-calling and agent states.

How to Build Effective Agentic Systems with LangGraph | Towards Data Science

Create AI workflows with agentic frameworks

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 20h

AI tools & vibe coding have hidden tradeoffs. Here's how to use them wisely and why open source is a great secret weapon.

By Clara Chong

How to Level Up Your Technical Skills in This AI Era | Towards Data Science

AI tools & vibe coding have hidden tradeoffs. Here's how to use them wisely and why open source is a great secret weapon.

towardsdatascience.com

1

Towards Data Science @towardsdatascience.com · 23h

Are you using #DeepSeek?

Matthew Gunton dives into how they formulated their policy optimization equations to allow for Reinforcement Learning.

Exploring DeepSeek's R1 Training Process | Towards Data Science

Open-Source Intelligence on Par with Proprietary Models

towardsdatascience.com

1

Towards Data Science @towardsdatascience.com · 1d

Mitchell Telatnik explores why traditional cyber risk models fall short and how applying some light statistical tools such as probabilistic modeling offers a better way forward.

Why Most Cyber Risk Models Fail Before They Begin | Towards Data Science

The case for quantitative thinking in cyber risk

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

The only way for businesses to keep up and move fast enough is with open source. Read @anacondainc.bsky.social Chief Product and Technology Officer Laura Sellers's full article FREE now.

Why Open Source is No Longer Optional — And How to Make it Work for Your Business | Towards Data Science

To stay ahead, open source is the only way.

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

Explore a complete, end-to-end workflow for temporal graph fraud detection. Erika G. Gonçalves walks through loading data, building a time-stamped graph, and preparing subgraphs for training.

No Peeking Ahead: Time-Aware Graph Fraud Detection | Towards Data Science

How to implement leak-free graph fraud detection

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

Understand how ROC curves and AUC help you go beyond accuracy with visuals and examples.

By Nikhil Dasari

ROC AUC Explained: A Beginner’s Guide to Evaluating Classification Models | Towards Data Science

Understand how ROC curves and AUC help you go beyond accuracy with visuals and examples.

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

In this follow-up piece, Murtaza Ali dives into visual encoding channels, offering a clear, non-technical look at how great data visualizations communicate meaning.

Data Visualization Explained (Part 2): An Introduction to Visual Variables | Towards Data Science

A non-technical and accessible guide to the underlying concept behind visual design: visual encoding channels

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

ICYMI: Thanh Liêm Nguyen's deep dive into applying causal inference to measure the effect of product unavailability on retail sales at Carrefour.

Analysis of Sales Shift in Retail with Causal Impact: A Case Study at Carrefour | Towards Data Science

Applying causal inference to measure the effect of product unavailability on retail sales at Carrefour

towardsdatascience.com

1 2

Towards Data Science @towardsdatascience.com · 1d

Your technical depth is what makes TDS unique. 💡

We're looking for your next practical guide — share what you've learned or struggled with this month.

👉 bit.ly/TDSContributor

1 2

Towards Data Science @towardsdatascience.com · 1d

Chinmay Kakatkar explores how graph algorithms can shape musical creativity, showing how Python can be used to compose melodies through Eulerian paths.

Eulerian Melodies: Graph Algorithms for Music Composition | Towards Data Science

Conceptual overview and an end-to-end Python implementation

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 1d

Maria Mouschoutzi explains how reranking in RAG pipelines helps surface the most relevant chunks, leading to more accurate and useful answers.

RAG Explained: Reranking for Better Answers | Towards Data Science

How reranking improves retrieval-augmented generation by surfacing the most relevant results

towardsdatascience.com

1

Towards Data Science @towardsdatascience.com · 2d

Sruly Rosenblat, Ilan Strauss, Isobel Moure, and Tim O’Reilly explore how Anthropic’s Model Context Protocol is reshaping the AI tool ecosystem. Mapping where power, access, and innovation now concentrate.

MCP in Practice | Towards Data Science

Mapping power, concentration, and usage in the emerging AI developer ecosystem

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 2d

Struggling with overly complex AI integrations? Discover LangGraph to simplify your development with agentic frameworks, focusing on what matters.

By Eivind Kjosbakken

How to Build Effective Agentic Systems with LangGraph | Towards Data Science

Create AI workflows with agentic frameworks

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 2d

Feel overwhelmed by complex data visualizations? Murtaza Ali sheds light on visual encoding channels, transforming raw data into easily digestible insights through effective use of visual variables like size and color.

Data Visualization Explained (Part 2): An Introduction to Visual Variables | Towards Data Science

A non-technical and accessible guide to the underlying concept behind visual design: visual encoding channels

towardsdatascience.com

Towards Data Science @towardsdatascience.com · 2d

Is your GPU starving? An inefficient data pipeline is often the hidden cause of slow model training, not the GPU itself. Andrea D’Agostino dives into the cascading bottleneck effect.

How to Improve the Efficiency of Your PyTorch Training Loop | Towards Data Science

Learn how to diagnose and resolve bottlenecks in PyTorch using the num_workers, pin_memory, and profiler parameters to maximize training performance.

towardsdatascience.com