650 Prod MLOps Case Studies Production is where machine learning meets business value, and Evidently AI has put together a comprehensive compendium of 650 real production ML/LLM case studies from 100+ companies (e.g., Netflix, Airbnb, DoorDash): It is great to see that the Evidently team continues to update and grow the list of production ML use-cases throughout the last few years, now also encompassing (of course) Generative AI and LLMs, as well as traditional ML such as computer vision, NLP, etc. There are also recurring use-cases across recommenders, search/ranking, and fraud detection; each including details on the in-house shipped system, covering product design, evaluation/metrics, and deployment/architecture. |
|
|
|---|
|
McKinsey on GenAI Minimal Value McKinsey has released a new report that calls out GenAI business value: "Nearly eight in ten companies have deployed gen AI in some form, but roughly the same percentage report no material impact on earnings.1 We call this the gen AI paradox.". As various other reports highlight, it seems most enterprises are stuck in perpetual "widespread adoption" with little to no P&L impact. Despite organisations scaling horizontal copilots, the high-value vertical use cases are stalling due to distraction towards the hype. The report is slightly ironic as it does seem to suggest that the fix to no value from GenAI is more GenAI, however it does provide sound suggestions on the importance of robust infrastructure to support rapid productionisation once value can be actually captured. There are call-outs to agentic AI mesh for orchestration, memory, tool adapters, and a shift from MLOps to AgentOps, however there is a significant portion on the opportunity building capabilities to capture value on the intersection of traditional ML and agents based systems. |
|
|
|---|
|
Google DeepMind Measuring Environmental Impact The environmental cost of a ChatGPT or Gemini prompt is larger than you can imagine; that's why Google had to publish their recent methodology to measure energy, emissions and water impact for Gemini products: Using this methodology Google estimates the median Gemini prompt uses 0.24 watt-hours of energy, emits 0.03 grams of carbon dioxide equivalent, and consumes 0.26 milliliters of water. These metrics have to be taken with a pinch of salt as they may be biased given the policy pressure towards the environmental impacts of these technologies. Having said that one thing that is clear is the efficiency improvements these models have been seeing throughout the recent past; in the report we can see improvements in Gemini of 33x/44x improvement, which does align with some of the huge investments we are seeing in research related to inference and training efficiency. |
|
|---|
|
Metadata Mgmt for AI Flows Metadata is the old-age proven and tested reliability layer for AI data pipelines, and this paper is a good reminder of how important it is in order to ensure robust production environments: This is an interesting paper which outlines TableVault as a Python-first metadata governance layer for human+LLM pipelines that makes every artifact, parameter, and operation auditable without replacing notebooks, ETL pipelines, or agents. It is interesting to see how concepts and methodologies are emerging to address the challenges in ML metadata management - particularly given that even outside of the agentic-contexts this is not a solved problem. This seems like a combination of taxonomy and tooling, however these tools are certainly not new, so it will be important to ensure that instead of re-inventing the wheel, these are able to ensure interoperability across the MLOps and DataOps tooling landscape. |
|
|
|---|
|
Beating Djikstra (for Sparse Graphs) A new shortest-path graph algorithm has emerged to take on good old Djikstra, however with the usual academic fine-print where in this case it's only relevant for sparse graphs: Having said that, as graph algorithms continue to become critical for every-day use-cases, an improvement on shortest-path algorithms could result in significant improvements across huge number of applications, particularly in the ML space. This paper presents a deterministic algorithm that apparently is able to circumvent the limitations presented by the sorting barrier in the Djikstra algorithm. It seems using a divide-and-conquer approach it is able to identify roots of large shortest-path subtrees and then recurse only on those, allowing for a theoretical O(m log^{2/3}n) bound algorithm. It is great to see how age-old algorithms still are able to see innovations from around the world. |
|
|
|---|
|
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:
Other upcoming MLOps conferences in 2025:
In case you missed our talks:
|
|
|---|
| | |
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! |
|
|---|
| | |
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!
|
|
|---|
| | |
| | | | The Institute for Ethical AI & Machine Learning is a European research centre that carries out world-class research into responsible machine learning. | | | | |
|
|
|---|
|
|
This email was sent to You received this email because you are registered with The Institute for Ethical AI & Machine Learning's newsletter "The Machine Learning Engineer"
|
| | | | |
|
|
|---|
|
© 2023 The Institute for Ethical AI & Machine Learning |
|
|---|
|
|
|
The Institute for Ethical AI & Machine Learning
