Issue #336 - The ML Engineer 🤖
Datadog OSS Observability ML, Anthropic Releases Claude 4, Upgraded Climate Foundation AI, Mistral Launches Devstral OSS, Microsoft AI Red Teaming Labs + more 🚀
Thank you for being part of over 70,000+ ML professionals and enthusiasts who receive weekly articles & tutorials on Machine Learning & MLOps 🤖 You can join the newsletter for free at https://ethical.institute/mle.html ⭐
If you like the content please support the newsletter by sharing with your friends via ✉️ Email, 🐦 Twitter, 💼 Linkedin and 📕 Facebook!
This week in Machine Learning:
Datadog OSS Observability ML
Anthropic Releases Claude 4
Upgraded Climate Foundation AI
Mistral Launches Devstral OSS
Microsoft AI Red Teaming Labs
Open Source ML Frameworks
Awesome AI Guidelines to check out this week
+ more 🚀
If you're looking for an interesting career opportunity, I'm hiring for a few roles including Data Science Manager (Forecasting), as well as Data Scientist (Forecasting) - check them out and please do share with your network!
In modern DevOps being able to predict failures before they happen can avoid huge outages and risks; Datadog out of the blue released a 151m+ param time-series foundation model for real time observability: Toto-Open-Base-1.0 is a large time-series foundation model specifically architected for the type of sparse, spiky, high-cardinality data in observability metrics. They also released "BOOM", a new (massive) benchmark of 350m real-world observations across 2,807 multivariate series. It is fascinating to see how time series foundation models have been evolving rapidly, and the datadog team are bringing innovations like patch-based causal normalization, proportional factorized attention, a Student-t mixture head, and a composite robust loss; of course the secret sauce is pretraining on over a trillion points of internal telemetry. The space of time series foundation models are really exciting, we are already seeing some (surprising) opportunities in their application in core business operations - certainly a key space to keep an eye on.
AI agents for software development are only growing in popularity, and Anthropic has just released Claude 4 with some exciting tooling to take the ecosystem to the next level: Claude 4 has been long awaited since the impact from 3.7, and this release comes with two models - Opus 4, which is optimized for sustained multi-hour agent workflows (72.5% on SWE-bench, 43.2% on Terminal-bench), and Sonnet 4, which balances efficiency for everyday use (72.7% SWE-bench). It is interesting also to see the extended support for reasoning that is doubling down on MCP/tool integration support. This also comes with Claude Code now moving to GA, and continuing to push the dev-agent race to see what further releases we'll continue to see week on week.
Upgraded Climate Foundation AI
Accurate forecasts of weather, air quality, and ocean hazards are critical both in environmental as well as business contexts; we see another exciting resource of Microsoft's foundation model for climate prediction: Microsoft's Aurora is a 1.3 billion-parameter foundation model that replaces multiple specialized Earth-system solvers with a unified encoder–processor–decoder architecture. It is interesting to see models that take a hybrid approach with deep learning with physics aware components, in this case it includes 1) a Perceiver-based 3D encoder ingests heterogeneous atmospheric and surface fields into a latent grid, 2) a multiscale 3D Swin Transformer U-Net evolves that state forward in time, and 3) a Perceiver-based decoder reconstructs physical variables at any resolution. This foundation model is pre-trained on over one million hours of mixed forecasts, reanalyses, and climate simulations for a next-step MAE objective - I continue to say this but this will be an exciting year for time-series foundation models across what seems to be a broad range of domains being disrupted.
Mistral enters the AI race for GenAI software agents with their new Devstral agentic LLM for software engineering: This new model is an exciting collaboration between Mistral AI and All Hands AI, and they explain it is designed to navigate large codebases, diagnose subtle bugs, and resolve real GitHub issues via agent scaffolds like OpenHands - . On the 500-issue SWE-Bench Verified benchmark it achieves 46.8% which suggests better performance than some of Anthropic's models, and is lightweight enough to run on a single RTX 4090 or a 32 GB-RAM Mac (with ~32 k token context). It will be quite interesting to see the results once the broader community is able to put it into practice - especially with the super recent Claude 4 release.
AI red teaming is only becoming more critical with the growing adoption of AI and the Microsoft AI team has released an full end-to-end Red Teaming Playground Lab: The Microsoft AI team originally developed a course for the “AI Red Teaming in Practice” course at Black Hat USA 2024 and they have now created an open source version available for free with12 interactive Jupyter-style challenges. This is a really great resource to dive into some GenAI security - it includes exercises from basic prompt-injection and metaprompt extraction attacks to multi-turn “Crescendo” exploits and guardrail bypass scenarios at increasing difficulty levels. Definitely worth checking out and getting hands on irrespective whether you are an ML practitioner, Software Developer, Security Practitioner or otherwise.
Upcoming MLOps Events
The MLOps ecosystem continues to grow at break-neck speeds, making it ever harder for us as practitioners to stay up to date with relevant developments. A fantsatic way to keep on-top of relevant resources is through the great community and events that the MLOps and Production ML ecosystem offers. This is the reason why we have started curating a list of upcoming events in the space, which are outlined below.
Upcoming conferences where we're speaking:
WeAreDevelopers 2025 - 9th July @ Berlin
SRE Con EMEA 2025 - 7th Oct @ Dublin
World Summit AI Europe - 08 Oct @ Amsterdam
Other upcoming MLOps conferences in 2025:
ODSC East - May 13 @ Boston
Data & AI Summit - 9th June @ San Francisco
AI Summit London - 12th June @ UK
MLOps World - Oct 8-9 @ Austin
In case you missed our talks:
Prod Generative AI in 2024 - KubeCon AI Day 2025
The State of AI in 2024 - WeAreDevelopers 2024
Responsible AI Workshop Keynote - NeurIPS 2021
Practical Guide to ML Explainability - PyCon London
ML Monitoring: Outliers, Drift, XAI - PyCon Keynote
Metadata for E2E MLOps - Kubecon NA 2022
ML Performance Evaluation at Scale - KubeCon Eur 2021
Industry Strength LLMs - PyData Global 2022
ML Security Workshop Keynote - NeurIPS 2022
Open Source MLOps Tools
Check out the fast-growing ecosystem of production ML tools & frameworks at the github repository which has reached over 10,000 ⭐ github stars. We are currently looking for more libraries to add - if you know of any that are not listed, please let us know or feel free to add a PR. Four featured libraries in the GPU acceleration space are outlined below.
Kompute - Blazing fast, lightweight and mobile phone-enabled GPU compute framework optimized for advanced data processing usecases.
CuPy - An implementation of NumPy-compatible multi-dimensional array on CUDA. CuPy consists of the core multi-dimensional array class, cupy.ndarray, and many functions on it.
Jax - Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
CuDF - Built based on the Apache Arrow columnar memory format, cuDF is a GPU DataFrame library for loading, joining, aggregating, filtering, and otherwise manipulating data.
If you know of any open source and open community events that are not listed do give us a heads up so we can add them!
OSS: Policy & Guidelines
As AI systems become more prevalent in society, we face bigger and tougher societal challenges. We have seen a large number of resources that aim to takle these challenges in the form of AI Guidelines, Principles, Ethics Frameworks, etc, however there are so many resources it is hard to navigate. Because of this we started an Open Source initiative that aims to map the ecosystem to make it simpler to navigate. You can find multiple principles in the repo - some examples include the following:
MLSecOps Top 10 Vulnerabilities - This is an initiative that aims to further the field of machine learning security by identifying the top 10 most common vulnerabiliites in the machine learning lifecycle as well as best practices.
AI & Machine Learning 8 principles for Responsible ML - The Institute for Ethical AI & Machine Learning has put together 8 principles for responsible machine learning that are to be adopted by individuals and delivery teams designing, building and operating machine learning systems.
An Evaluation of Guidelines - The Ethics of Ethics; A research paper that analyses multiple Ethics principles.
ACM's Code of Ethics and Professional Conduct - This is the code of ethics that has been put together in 1992 by the Association for Computer Machinery and updated in 2018.
If you know of any guidelines that are not in the "Awesome AI Guidelines" list, please do give us a heads up or feel free to add a pull request!
About us
The Institute for Ethical AI & Machine Learning is a European research centre that carries out world-class research into responsible machine learning.