💬 We’d love your feedback! At Kedro, we’re always looking to improve and better serve our community. That’s why we’re running our Net Promoter Score (NPS) survey — a quick way for you to share how likely you are to recommend Kedro and tell us what’s working (and what’s not). 👉 Your input directly shapes how we prioritise improvements and build the Kedro you want to see. It only takes 1 minute to complete, and your feedback will make a real impact: https://www.surveys.online/jfe/form/SV_0HvXz6XqDqaP8ou Thank you for helping us grow Kedro together 🚀

<!here> ☝️ Please help us make Kedro better 🙂

In the next edition of the ☕ Kedro Coffee Chat 🔶, we’ll explore SparkDatasetV2 and modern Databricks workflows with Kedro. @Sajid Alam will walk us through the SparkDatasetV2 rewrite — why it was needed, what problems it solves, and how it simplifies working with Spark datasets in Kedro. We’ll also cover the three supported ways to run Kedro on Databricks today: Git folders, Databricks Connect for local-first development, and production deployments via kedro-databricks. And we’ll finish with a live demo. 📅 When: 13 March, 1 PM (GMT+0) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/posts/in-this-edition-of-kedro-coffee-chat-ugcPost-743754185[…]m=member_desktop&amp;rcm=ACoAADMeBAABWge3NEpWk5ncE7V4ptixUfT_f9c 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

We’re starting ☕ Kedro Coffee Chat 🔶 soon!

📣 🔶 Quick reminder to fill in our SHORT survey, it'll take less than 1 minute. Survey closes in 7 days!

Hi Kedro community, please take a moment to fill in the survey above 👆 it only takes a minute, and your feedback is extremely valuable to us!!! 🙏K

<!here> Kedro friends, don't forget to take 1 min to fill in the survey here👇, your feedback is extremely valuable to us in shaping Kedro as a better product! Only 5 days left! 🙏K survey: https://www.surveys.online/jfe/form/SV_0HvXz6XqDqaP8ou

Let AI Agents Manage Your ML & Data Pipelines with Kedro and MCP In the next edition of the ☕ Kedro Coffee Chat 🔶, we’ll explore how AI agents can manage ML and data pipelines with Kedro and MCP. @Dmitry Sorokin will show how Kedro pipelines can be wrapped with an MCP server, allowing AI agents built with LangGraph or Claude Code to interact with and orchestrate workflows programmatically. This approach combines the deterministic nature of production-ready pipelines with the flexibility and reasoning capabilities of modern AI systems. 📅 When: 27 March, 1 PM (GMT+0) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/posts/in-this-edition-of-kedro-coffee-chat-ugcPost-744256840[…]m=member_desktop&amp;rcm=ACoAADMeBAABWge3NEpWk5ncE7V4ptixUfT_f9c 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

Ship GenAI with Confidence: Evaluation Pipelines with Kedro and Langfuse In the next edition of the ☕ Kedro Coffee Chat 🔶, we’ll explore how to bring systematic evaluation into production GenAI workflows using Kedro and Langfuse. We’ll introduce

LangfuseEvaluationDataset

— a recently released dataset that makes evaluation a first-class part of your Kedro project. We’ll demo how to define test cases in configuration, run experiments across different prompt versions and models, and compare results side-by-side in the Langfuse UI. 📅 When: 10 Apr, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7447960424517582848?viewAsMember=true 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

KKK Kedro 1.3.1, Kedro-Datasets 9.3.0, and VSCode-Kedro 0.7.0 are out! KKK Hi everyone, we're excited to announce three new releases! 🥳🥳🥳 • Kedro 1.3.1 brings optional parameter validation using type hints, letting you auto-validate and instantiate Pydantic models and dataclasses with zero impact on untyped parameters. • Kedro-Datasets 9.3.0 adds pandas 3.0 compatibility and new Ibis backends. • VSCode-Kedro 0.7.0 delivers a much-improved developer experience with auto-reload, theme toggling, and multi-project support. 🚀 What can you expect in these releases? K Kedro 1.3.1 • Major features and improvements: ◦ Optional parameter validation with Pydantic model and dataclass support via type hints. ◦ New

list_versions()

method for versioned datasets. ◦

pipelines_to_find

parameter in

find_pipelines()

for selectively running pipeline subsets. ◦

--checkout

flag now works on new projects without a starter. ◦

SESSION_CLASS

configurable project setting for custom

KedroSession

subclasses. ◦ Improvements to error messages, logging configuration, and long-running server support read our release notes for more information. • 📚 Documentation changes: ◦ Added parameter validation docs covering Pydantic model and dataclass support for typed parameters. ──────────────────────── K Kedro-Datasets 9.3.0 • Major features and improvements: ◦ Compatible with pandas 3.0. ◦ New

ibis-materialize

and

ibis-singlestoredb

extras for Ibis 12.0 backends. ◦ "Upsert" save mode added to

ibis.TableDataset

via

MERGE INTO

support. ◦ New experimental

langfuse.LangfuseEvaluationDataset

for managing Langfuse evaluation datasets (

kedro_datasets_experimental.langfuse

). ◦ Additional fixes and compatibility improvements across datasets read our release notes for more information. ──────────────────────── K VSCode-Kedro 0.7.0 • Major features and improvements: ◦ Auto-reload Kedro-Viz on project file changes. ◦ Toggle Kedro-Viz theme directly from VS Code. ◦ Migrated Webview from

react-scripts

to Vite. ◦ Relative path support in

kedroProjectPath

, auto-detection of Kedro projects, and multiple project support. ◦ Developer experience, security, and dependency updates read our release notes for more information. ──────────────────────── 🙌 Community contributions: A huge thank you to everyone who contributed to these releases! Kedro 1.3.1: aziq, zhubaobao2024, Camille Coeurjoly, sinanpl, Mr-Neutr0n, mvhensbergen, SayantanDutt Kedro-Datasets 9.3.0: Priyanka, Akumawavez, Joris, Bas-commits, oomenn, Celina, Juanchodpg2 VSCode-Kedro 0.7.0: pascalwhoop 🎯 How to get the latest releases • Python users

pip install kedro==1.3.1
pip install kedro-datasets==9.3.0

• VS Code users

Search for Kedro in the Extensions marketplace and update to v0.7.0.

🔗 Helpful links • Kedro-datasets GitHub repository • Kedro core GitHub repository • VSCode-Kedro GitHub repository • Kedro docs You too can contribute by using these features and providing feedback. Follow our roadmap for more updates, comment on the tickets, and suggest future features to the community. Until next time, The Kedro Team 💛

Now serving: Exposing Kedro Pipelines over HTTP In the next edition of ☕ Kedro Coffee Chat 🔶, we'll be introducing a service-oriented API layer and all new KedroServiceSession that enables multi-run workflows, seamless external integration, and the ability to treat Kedro pipelines as reusable services. 📅 When: 24 Apr, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7453023045264248832/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

<!here> Happening now! ☝️

Hi everyone 👋 We’ve built a visual tool to create Kedro pipelines—no CLI needed—and would love your feedback! 👉 Try it here: demo.kedro.org/kedro-builder What it is A drag-and-drop interface where you can: • Create Kedro projects visually • Add datasets and function nodes • Connect components into pipelines • View the generated code • Export a ready-to-use Kedro project Great for • Quick prototyping • Understanding how datasets, nodes, and pipelines fit together How to try it 1️⃣ Follow a quick flow: • Create a new project • Add a dataset (CSV, Parquet, etc.) • Add a function node • Connect them into a simple pipeline • View the generated code • Export your project 2️⃣ Or explore freely: • Try the 42 supported dataset types • Build more complex pipelines • Test validation and error handling • Inspect the generated Kedro code While testing, we’d love your thoughts on: • Was it intuitive to use? • Did the generated code look correct? • Would you use this for learning or prototyping? • Was there anything that felt confusing or unclear? • Is there anything you’d like to see added or improved? You can share feedback via feedback button in the tool or DM me directly. Thanks for helping make Kedro more accessible 🙏

Spec-Driven Development for Data Practitioners In the next edition of ☕ Kedro Coffee Chat 🔶, @Rashida Kanchwala will explore spec-driven development and how OpenSpec can be combined with Kedro to build reliable, well-defined data workflows. 📅 When: 08 May, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7458142146634698752/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

<!here> 🗞️ Hot off the press: Generative AI workflows need engineering discipline to scale beyond the demo It's been less than a year since we released Kedro 1.0 K, and a lot has changed in how we think about and build data pipelines. Agentic workflows have moved past the novelty stage and teams are now implementing them alongside traditional ML steps. We've shipped a number of features across the Kedro ecosystem to meet these evolving needs: the LLM context node for preparing everything required to run an LLM, datasets for configuring and managing prompts, tracing, and evaluation. Now the team has published an article sharing our vision for how Kedro fits into the agentic workflow, bringing engineering discipline to ensure these systems don't stay as demos but actually make it to production! You can find the article here: https://medium.com/quantumblack/generative-ai-workflows-need-engineering-discipline-to-scale-beyond-the-demo-a9928135e393 🔗

Should AI Write Your Docs? A Team Discussion ☕ In the next edition of ☕ Kedro Coffee Chat 🔶, @Laura Couto will explore the use of AI for writing documentation: what works, what doesn't, and where the line should be. Building on a recent team discussion, she'll share pros and cons from real experience. 📅 When: 22 May, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7463191435249561600/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

KKK Kedro 1.4.0, Kedro-Datasets 9.4.0, and Kedro-Telemetry 0.8.0 are out! KKK Hi <!here>, we’re excited to announce three new releases! 🥳🥳🥳 • Kedro 1.4.0 brings official python 3.14 support, all new

KedroServiceSession

, and project inspection API • Kedro-Datasets 9.4.0 adds python 3.14 support, several new

huggingface

datasets as well as the

opik.EvaluationDataset

as an experimental dataset • Kedro-Telemetry 0.8.0 also adds python 3.14 support and fixes bugs in the dataset type collection 🚀 What can you expect in these releases? K Kedro 1.4.0 • Major features and improvements: ◦ Official support for Python 3.14 ◦ First version of the

KedroServiceSession

which enables runtime parameter injection per run as well as multiple runs per session ◦ Ability to inspect the project snapshot via the inspection API ◦ Improved

%load_node

and scoped parameter type-hint validation when selected pipelines are being run ◦ For more changes, see the full release notes • 📚 Documentation changes: ◦ Added documentation for

KedroServiceSession

◦ Fixed API docs the

session

and

context

modules ──────────────────────── K Kedro-Datasets 9.4.0 • Major features and improvements: ◦ New HuggingFace Datasets:

ArrowDataset

,

ParquetDataset

,

JSONDataset

and

CSVDataset

◦ Support for Python 3.14 (except for

tensorflow.TensorFlowModelDataset

and

geopandas.GenericDataset

due to no support in underlying libraries) ◦ Credentials support in

ibis.FileDataset

◦ New experimental dataset:

opik.EvaluationDataset

◦ Some experimental datasets for

langfuse

,

opik

and

langchain

have been renamed to simplify the naming convention ◦ See full release notes for more details ──────────────────────── K Kedro-Telemetry 0.8.0 • Major features and improvements: ◦ Support for Python 3.14 ◦ Fixed bug related to collection of dataset types ──────────────────────── 🙌 Community contributions: A huge thank you to everyone who contributed to these releases! Kedro 1.4.0: BhavayChopra, jeevan6996, Rahul Bansod, PragnyaKhandelwal, datascienceio, Guillaume Tauzin, Prince Kumar Kedro-Datasets 9.4.0: datascienceio, Guillaume Tauzin, Ian Whalen, Sai Asish Y, Kaushal Dhungel, Anton Nikishin 🔗 Helpful links • Kedro-datasets GitHub repository • Kedro core GitHub repository • VSCode-Kedro GitHub repository • Kedro docs You too can contribute by using these features and providing feedback. Follow our roadmap for more updates, comment on the tickets, and suggest future features to the community. Until next time, The Kedro Team 💛

<!here> 🎉 Kedro went open source 7 years ago today, back in 2019! Happy anniversary to everyone who has been part of the journey - our users, contributors, maintainers, and advocates. Thank you for helping build, improve, and grow the Kedro ecosystem over the years. Here's to many more years of better data and AI engineering together! 🎂🚀

KKK Kedro-Viz 12.4.0, and VSCode-Kedro 0.8.0 are out! KKK Hi everyone, we’re excited to announce two new releases! 🥳🥳🥳 • Kedro-Viz 12.4.0 brings official Python 3.14 support, and expose full internal Kedro node name as

kedro_node_name

. • VSCode-Kedro 0.8.0 adds a native VS Code right-click action on Kedro Viz task nodes to open an untitled debug notebook prefilled with

%load_node <node_name>

. 🚀 What can you expect in these releases? K Kedro-Viz 12.4.0 • Major features and improvements: ◦ Official support for Python 3.14 ◦ Expose full internal Kedro node name as

kedro_node_name

◦ Add

NodeContextMenu

support for VSCode ◦ Add Consent Management Platform (CMP) script to Kedro-Viz docs ◦ For more changes, see the full release notes ──────────────────────── K VSCode-Kedro 0.8.0 • Major features and improvements: ◦ Adds a native VSCode right-click action on Kedro Viz task nodes to open an untitled debug notebook prefilled with

%load_node <node_name>

. ◦ For more changes, see the full release notes 🙌 Community contributions: A huge thank you to everyone who contributed to these releases! Kedro-Viz 12.4.0: Eoin Shaughnessy and tuanaiseo 🔗 Helpful links • Kedro-Viz GitHub repository • VSCode-Kedro GitHub repository You too can contribute by using these features and providing feedback. Follow our roadmap for more updates, comment on the tickets, and suggest future features to the community. Until next time, The Kedro Team 💛

Kedro in VS Code: From Pipeline View to One-Click Node Debugging ☕ In the next edition of ☕ Kedro Coffee Chat 🔶*,* @Jitendra Gundaniya will be introducing the Kedro VS Code extension and a new one-click node debugging workflow. We will look at how developers can move from a visual Kedro-Viz pipeline inside VS Code to an interactive notebook prepared with

%load_node

, making it easier to inspect and debug individual Kedro nodes. 📅 When: 05 June, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7468274192220655616/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

Today marks the 7th anniversary Kedro goes open source. K🎂 They say 7 years represents a full cycle, a time to reflect on our journey and look forward to the next major milestone. Kudos to everyone here who contributes to and supports this open-source community. Here’s to the next 7 years! 🪅

GraphRAG with Kedro ☕ In the next edition of ☕ Kedro Coffee Chat 🔶, @Laura Couto will demo a complete GraphRAG system built entirely with Kedro. Starting from 55,500 synthetic patient records, she'll walk through five modular pipelines that clean the data, build a knowledge graph, index it into a vector database, and power an agentic Q&A interface all with every dataset tracked in the catalog. If you've been curious about how Kedro handles GenAI workflows end-to-end. 📅 When: 19 June, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7473343901513326593/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

<!here> You don't want to miss this one!! ☕ ☝️

KKK Kedro 1.5.0 and Kedro-Datasets 9.5.0 are out! KKK Hi everyone, we’re excited to announce a new Kedro release! 🥳🥳🥳 Kedro 1.5.0 introduces a brand new Server API layer powered by

KedroServiceSession

, a

kedro server start

CLI command, selective pipeline loading, and updated deployment documentation. announcement Read all about Kedro Server API and KedroServiceSession here! https://kedro.org/blog/kedro-as-a-service Kedro-datasets 9.5.0 introduces a security fix to

PyTorchDataset

,

send_individually

option to

APIDataset

and other minor improvements. 🚀 What can you expect in this release? K Kedro K Major features and improvements: • New HTTP Server API layer built on

KedroServiceSession

to execute pipelines from HTTP requests, with endpoints for: -

/run

— run pipelines with optional runtime parameters -

/health

— check the server’s health status -

/snapshot

— retrieve the current project snapshot • New CLI command

kedro server start

to launch the server • Selective pipeline loading: when

--pipelines

is specified, only the requested pipeline modules are imported, speeding up startup • Refreshed the Amazon EMR Serverless, AWS Step Functions, and AWS Batch deployment guides with strategy-first layouts, namespace-based grouping, container image guidance, and end-to-end walkthroughs • And more! Read the release notes for a full list of changes! K Kedro Datasets K • Added `send_individually`option to`APIDataset`to send list items as individual requests instead of batched arrays. • Hardened experimental `pytorch.PyTorchDataset`:`weights_only=True`is now enforced by default on load to block arbitrary code execution from untrusted`.pt`files, user-supplied`load_args`and`save_args`are now correctly passed to`torch.load`and`torch.save`(previously silently dropped), and the misleading “pickle-safe” docstring was corrected. • Fixed`ibis.FileDataset`to support remote filepaths (e.g.

s3://

,

abfss://

,

hf://

) and added an`fs_args`argument to authenticate the filesystem used for version discovery. • And more! Read the release notes for a full list of changes! 🙌 Community contributions: A huge thank you to everyone who contributed to this release: PragnyaKhandelwal, Anton Nikishin, GDaamn, Feng Jikui, Rudra Dudhat, Jean-Baptiste Braun, Gargi, and Simon 🔗 Helpful links • Kedro GitHub repository • Kedro-datasets GitHub repository • <http://(http://docs.kedro.org|Kedro docs> You too can contribute by using these features and providing feedback. Follow our roadmap for more updates, comment on the tickets, and suggest future features to the community. Until next time, The Kedro Team 💛

<!here> We're not bringing one, but two releases across the Kedro ecosystem today! K ☝️

All about the Data Catalog ☕ In the next edition of ☕ Kedro Coffee Chat 🔶, @Merel goes all in on the Data Catalog, Kedro's single source of truth for your data. She'll walk through the catalog fundamentals, configuration environments, configuration patterns, templating, and Dataset Factories, which let you swap many repetitive catalog entries for one clean pattern. Whether you're brand new to the catalog or want to stop hardcoding filepaths for good, come hang out. 📅 When: 03 July, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: https://www.linkedin.com/events/7478403211779870720/ 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

<!here> Looking forward to telling you all about Kedro's power feature: Data Catalog ☝️

Hi everyone! We’ve recently released K Kedro 1.5.0 K with some exciting new features - the Kedro Server API and the new KedroServiceSession! 📣 Read the full blog post here: kedro.org/blog/kedro-as-a-service We’ve been working on enabling “Kedro as a service” - making Kedro pipelines callable within long-running processes and over HTTP. This unlocks interactive apps, repeated-execution loops, API driven workflows and integration with AI agentic systems. The two new pieces are now available to try: 🌐 Kedro Server API — a FastAPI server with

GET /health

,

POST /run

, and

GET /snapshot

uv pip install 'kedro[server]'
kedro server start --host 127.0.0.1 --port 8000

🔁 KedroSeviceSession — run pipelines repeatedly in one process; state loads once, each run stays isolated:

with KedroServiceSession.create() as session:
    session.run(runtime_params={"param1": "value1"})
    session.run(runtime_params={"param1": "value2"})

We’d love your feedback! How would you use Kedro as a service? What’s missing for your use case? Drop a comment here or open an issue/discussion on GitHub. 🙏

Reflection Hub ☕ In the next edition of ☕ Kedro Coffee Chat 🔶, @Ravi Kumar Pilla will introduce Reflection Hub which is a system for building reusable reflection and learning capabilities into enterprise AI agents and workflows. If you're interested in making your AI workflows smarter with each run, this is one you won't want to miss. 📅 When: 17 July, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: linkedin.com/events/7483455868638814208 📣 These sessions are public and open to everyone. Join us live or catch the recording afterwards.

In the next edition of ☕ Kedro Coffee Chat 🔶, @Sajid Alam will preview dataset validation, a new capability that catches bad data at the I/O boundary before it ever reaches your nodes. Declare a validator on a catalog entry with one line of YAML, and Kedro validates every load and save in pipeline runs, notebooks and CI, reporting every failed check at once instead of a mystery stack trace deep inside a node. There will be a live demo on a spaceflights project, a peek at how it works under the hood, and the roadmap to release. 📅 When: 07 August, 1 PM (GMT+1) youtube YouTube: linkedin LinkedIn: linkedin.com/posts/in-this-edition-of-kedro-coffee-chat…?rcm=… 📣 These sessions are public and open to everyone. Join us live or catch the recording.