By no means miss a brand new version of The Variable, our weekly e-newsletter that includes a top-notch number of editors’ picks, deep dives, neighborhood information, and extra.
May it actually be the top of one other 12 months? We’ve been publishing annual recaps for a very long time, and this closing stretch nonetheless one way or the other sneaks up on us.
Yearly feels hectic, after all. However we’re fairly positive that we’ve by no means been busier — in the very best methods! — than in 2025, whether or not it’s rebuilding TDS as an independent publication, iterating on our Author Payment Program, or staying as decided as ever to publish the best articles by data science, ML, and AI experts.
On the finish of such a monumental 12 months, we’re thrilled to spotlight the tales that stood out among the many a whole lot we’ve printed, and our celebration of 2025 must-reads displays the range of subjects and experiences our authors cowl.
Yearly, a recent, thrilling set of instruments inevitably captures practitioners’ creativeness; after LLMs in 2023 and RAG in 2024, this has been, with out a query, the Year of the Agent (and complementary frameworks like MCP and contextual engineering).
Past agentic AI, Python continued its long reign as a necessary programming language, and our authors and readers alike zoomed in on maintaining their expertise up-to-date in a aggressive job market.
Earlier than we discover the articles that resonated essentially the most with our readers this 12 months, we’d wish to take a second to thanks for trusting us along with your curiosity, and our authors for becoming a member of us in our new house.
Brokers, Brokers, and Extra Brokers
Whereas agentic AI wasn’t fairly new this 12 months — it already popped up as a trending matter in our previous annual recap! — its attain and mainstream standing grew exponentially. Listed here are the highest tales on the signature know-how of 2025.
Methods to Design My First AI Agent
In 2025, everyone working in, round, or with AI wished to know what brokers are — and learn how to leverage their energy. Fabiana Clemente’s blockbuster tutorial responded to this want with clear, actionable pointers that may be personalized for particular contexts and use instances.
A Developer’s Information to Constructing Scalable AI: Workflows vs Brokers
A masterful deep dive from Hailey Quach, who unpacks the tradeoffs between autonomous brokers and orchestrated workflows and the conditions during which brokers are referred to as for.
LangGraph 101: Let’s Construct A Deep Analysis Agent
As Shuai Guo explains, “Constructing LLM brokers that truly work in observe isn’t a straightforward process” — which explains why an accessible information like this one resonated with so many readers.
Agentic AI: Single vs Multi-Agent Methods
AI brokers are available in numerous flavors, and we are able to use them for a variety of duties. Ida Silfverskiöld focuses on one key distinction to be conscious of.
AI Brokers from Scratch: Single Brokers
Relying solely on Python and Ollama, Mauro Di Pietro shared one among final 12 months’s standout hands-on tutorials, exhibiting us learn how to construct a useful agent.
Should-Have Expertise, Profession Progress, and Different Trending Matters
From the newest buzz-generating ideas to the evergreen areas that preserve information and ML professionals aggressive in a difficult job panorama, our high authors lined a powerful vary of subjects and questions.
Methods to Grow to be a Machine Studying Engineer (Step-by-Step), by Egor Howell
Our most-read article of 2025 presents a one-stop information to changing into a machine studying engineer.
I Gained $10,000 in a Machine Studying Competitors — Right here’s My Full Technique, by Claudia Ng
Full information to characteristic choice, threshold optimization, and neural community structure for ML competitions.
Superior Immediate Engineering for Information Science Initiatives, by Sara Nobrega
Partly 2 of Sara’s common collection, we study all about immediate engineering for options, modeling, and analysis.
Is RAG Lifeless? The Rise of Context Engineering and Semantic Layers for Agentic AI, by Steve Hedden
How retrieval-augmented technology is evolving from static pipelines to ruled, context-aware techniques that make AI extra explainable, reliable, and scalable.
Mannequin Context Protocol (MCP) Tutorial: Construct Your First MCP Server in 6 Steps, by Destin Gong
A step-by-step information to develop a customized code-to-diagram MCP server.
Python, Without end
Tendencies and buzzwords come and go, however the potential to code in Python is as related as ever. Listed here are our high Python-focused reads of the previous 12 months.
Constructing A Fashionable Dashboard with Python and Taipy, by Thomas Reid
From an writer who’s printed many a viral Python article on TDS (be sure you revisit a few of Tom’s different tales!) comes a information to constructing a front-end information software.
How We Lowered LLM Prices by 90% with 5 Traces of Code, by Uri Peled
When clear code hides inefficiencies: what we discovered from fixing a number of traces of code and saving 90% in LLM price.
Implementing the Espresso Machine in Python, by Mahnoor Javed
A beginner-friendly, step-by-step information to coding a espresso maker (and studying about conditional statements, loops, and Python dictionaries alongside the best way).
Contribute to TDS
Serious about drafting a brand new article over the vacations? We’d like to learn it. Don’t hesitate to send it our way.
