Licence: CC BY-NC-ND 4.0 MLOps : Inference Architecture Pattern Context Natural Language Processing (NLP) are dedicated AI application that works on text. This application use Large Language Model (LLM) with billions of parameters build with Deep Learning technics. To compute efficiently a vector representation of text, this model must be loaded and executed in Graphical Processing Unit (GPU) most of the time. This GPU card are expensive, more that 1 euro per hour on AWS Cloud and LLM not optimized most of the time....
🧠 AI Confidential 😎
Compute your predictions without any data transfer Data availability is one of the strongest limitations for the usage of machine learning algorithms today in many organisations. How many hours of meetings and effort are required to resolve a data governance issue or assess compliance with data management policies between your different units, etc … It is time to change the paradigm, move the computation not the data. Indeed, it is now possible to load in your user’s browser a complete Machine Learning application....
🐍 Unix socket interaction with python
Today it is increasingly easy to integrate third-party components into your software. However, including new dependencies is also risky for your business. This is why you must regularly justify the introduction of a new dependency to your team. We have recently seen with the implementation of multiple factor authentication on the python package repository issue. When the maintainer of the package atomicwrite remove it to protest against a good decision to improve the software supply chain security....
🕵🏻♂️ 🧾 Tinycheck with a receipt printer
Tinycheck is an open source stalkerware detection tool mainly developped by Felix Aimé It based on a device to which your phone will connect via the wi-fi network. During your navigation it will record a network trace. That trace will then be analyzed to detect stalkerware on your phone. For a good and complete description, I suggest to read that arcticle : TinyCheck: Stalkerware detection that doesn’t leave a trace from MalwareBytes....
🕸 Hybrid cloud with a reverse and a dynamic proxy
When implementing new machine learning algorithms, it may be useful to go through an experimentation phase during which your algorithm is made available to users for testing without necessarily requiring a production and exepensive environment with horizontal scaling on the cloud. DALL-E Mini :AI image generated from text : A photorealistic a cable-stayed bridge to the cloud in a sunny sky- Indeed, the very high cost of the cloud pushes to the implementation of hybrid architecture scenario for which we will use constant but cheap local resources to validate a concept, an idea or an algorithm....
Helloword
👋