Towards Data Science

The world's leading publication for data science and artificial intelligence professionals.

Submit an Article ✍️

[Towards Data Science @TDataScience]

We're thrilled to welcome Anand Rawat to TDS — we invite everyone to explore his excellent article on transforming long-running Python simulations into a self-service platform. https://t.co/9W3u4szr6z

[Towards Data Science @TDataScience]

Don't miss the latest installment in @ibbysalam's data engineering learning journey — this time, walking us through the process of building an incremental pipeline with dbt snapshots. https://t.co/vRSFBZdxmN

[Towards Data Science @TDataScience]

For an incisive look at the areas where AI models hit a wall, read Lars Benedikt Kaesberg and Jan Philip Wahle's debut TDS article, which focuses on spatial reasoning and the grid puzzles that trip up powerful language and vision models. https://t.co/UwIHWKjSL5

[Chien Vu @chien_vu1692]

Cheaper LLMs ≠ Lower costs.

Great insights from Nhu Hoang on @TDataScience about the hidden costs of LLM routing.

Optimizing AI isn't just about model pricing—it's about total cost, latency, and quality.

Worth a read 👇

https://t.co/LEKTGBB45e

#LLM #AIEngineering

[🍞 @chickenlearning]

Okay, finally I can announce it.

So excited to make a comeback for @TDataScience !

So here is the glimpse for what I made,

[Lars Kaesberg @lbkaesberg]

I wrote an article for @TDataScience, and it ended up as an Editor’s Pick

It’s about grid puzzles people solve easily, but the best AI models still get stuck on, and what this tells us about their spatial reasoning skills. Thanks to @jpwahle for the help!
https://t.co/Zs1TdEWRsU

[Nhu Hoang @nhu_hoang]

Cheaper LLMs don't always mean lower costs. While exploring LLM routing, I found that switching models has hidden costs too. Check it out at @TDataScience
#LLM #GenerativeAI #AIEngineering #LLMRouting #AgenticAI

[Himanshu Sharma @Himansh70809561]

Stacking doesn't automatically beat your best model.

Diverse models: 0.830 vs 0.820 ✅
Correlated tree ensembles: +0.1 at 35x the training cost ❌
Plain voting lost to the best single model both times.
@TDataScience
Full test + notebook:
https://t.co/YidgPctmGw

[Towards Data Science @TDataScience]

Behind every news story about an AI-related safety failure, we can likely detect a range of technical causes.

Stephanie Kirmer unpacks the processes that can lead to such failures and the steps teams and practitioners can take to mitigate them. https://t.co/1rWfvZEYuG

[Towards Data Science @TDataScience]

"Every additional EEG channel adds another electrode, another signal stream, another source of setup complexity, and another feature block that must be processed at inference time."

Renish Joshi looks at various approaches to making EEG models perform more efficiently. https://t.co/q5l1RcBv38

[Towards Data Science @TDataScience]

What happens to EEG models when they "see" the same people in both training and testing sets? Renish Joshi introduces us to the challenges of participant-aware evaluation. https://t.co/XH4kmiCcgh

[Towards Data Science @TDataScience]

What does it take to push a ML model from accurate to reliable? Devashish Thakar walks us through one approach leveraging the conformal prediction framework. https://t.co/kMLbnP2eig

[Towards Data Science @TDataScience]

Threat modeling where the LLM can't invent a CVE, and code computes the risk score

By Daniel Nweke

https://t.co/UJ4VNhJJWt

[Towards Data Science @TDataScience]

From benchmarks and current pricing to its underlying architecture, @BenjaminNweke11 recently shared a helpful introduction to Gemini 4 Argon. https://t.co/xMqF1GObzL

[Towards Data Science @TDataScience]

"Rather than repeat the theory from my earlier article, I wanted to ask a more useful question: what happens if I run exactly the same code on both versions?"

@taupirho takes the Python 3.15 JIT for a spin. https://t.co/TtJnuFBckB

[Towards Data Science @TDataScience]

What changes once an LLM comes with a million-token output window?

@BenjaminNweke11 walks us through the nitty-gritty details of using Google's new Gemini 4 Argon model. https://t.co/xMqF1GObzL

[Towards Data Science @TDataScience]

For his debut TDS article, Devashish Thakar presents an accessible deep dive on conformal prediction and its practical use cases. https://t.co/kMLbnP2eig

[Towards Data Science @TDataScience]

In a thorough and engaging walkthrough of a recent computer vision project, Florian Trautweiler unpacks the process behind synthetic-to-real transfer leveraging frozen DINOv3 features. https://t.co/Bd1VGdExap

[Towards Data Science @TDataScience]

"A better model follows advice more often but not always. Setting the temperature to zero makes its answers more repeatable, not more obedient."

In the latest installment of his Towards Spec-Driven Test Automation series, @GA4198498563411 weighs the benefits and drawbacks of using skills vs. an MCP server. https://t.co/xxlbLA5vZu

[Towards Data Science @TDataScience]

Switching to a cheaper model should... lower your overall costs, right?

@nhu_hoang Nhu Hoang explains why that's not always the case, and stresses the importance of verifying the hidden costs of LLM routing. https://t.co/9hnP91ETBI