Jueves de Quack con Bruno Capuano

Introduction and Host Introduction

• The host, Bruno, introduces himself and welcomes viewers to the episode, mentioning that he is filling in for Andrea, and he works at GitHub/Microsoft, living in Canada but originally from Argentina 10s.
• Bruno explains that he will be discussing students, technology, and other topics with his guests, Francis and Cristian, and invites viewers to share their locations in the comments, while also mentioning his work with artificial intelligence and programming languages such as Python, C#arp, and C++ 1m42s.

Guest Introductions

• Francis, one of the guests, introduces himself as a student of systems engineering from Cochabamba, Bolivia, and mentions that he is an ambassador for ITO Senior, previously known as ITO Gold, and is nearing the end of his studies 4m6s.
• Cristian, the other guest, also from Cochabamba, Bolivia, introduces himself and thanks the host for the invitation, mentioning that he is close to finishing his computer science degree and is involved in the Microsoft Ambassadors program, as well as being the founder of some communities in his hometown 5m30s.

Introduction to the Discussion and Upcoming Trip

• The introduction to the discussion is made, thanking everyone and acknowledging the professionalism of Cristian, and then moving on to talk about an upcoming trip to Spain, where various events will be attended, 10s.
• A personal connection to Spain is shared, mentioning that it is one of three places in the world where the heart is, along with Argentina and Canada, and that it is a great place to visit due to its beautiful people, good food, and quality of life, 1m42s.

Introduction to Microsoft Build Event

• The topic of Microsoft Build is introduced, explaining that it is one of the major events of the year, focusing on creators and featuring a wide range of content for programmers, including topics such as data, cloud, security, and more, 2m6s.
• The details of the Microsoft Build event are shared, including its location in San Francisco, the option to attend in-person or register online for free to view sessions, and the variety of sessions available, including a keynote by Satia, the CEO of Microsoft, 3m30s.
• The catalog of sessions for Microsoft Build has been published, and it is possible to register on the website, which is bill.microsoft.com, where attendees can expect to find a wide range of sessions and labs, including digital labs that provide a voucher for access to a virtual environment to learn from experts, 5m10s.

Introduction to AI Command-Line Tools and GitHub Copilot

• A tool is available that allows users to interact with AI models using a command-line interface, which is a natural way to communicate with these tools, and this method is particularly useful for those who use EAI, such as GitHub Copilot, 10s.
• The most popular and well-known method of interacting with AI is through MCPs, and skills have also been introduced, with GitHub Copilot now supporting skills, but using the command line is still a simple and natural way for AI models to work, 10s.
• A plugin is available for GitHub Copilot that allows users to install and use a command-line interface, and this plugin is fascinating because it enables users to work with AI models in a straightforward way, 42s.

Olama Tool and Its Features

• The plugin can be installed in GitHub Copilot, and once installed, users can work with various projects, such as the Olama Monitor, which is a small tool that runs on Windows and provides information on how Olama is functioning, 2m6s.
• Olama is a tool that allows users to run local models, including popular models like Quen 36, Nemotron, and Gema, and it can be installed on Windows, Linux, and Mac, with a simple installation process, 2m6s.
• Once Olama is installed, it appears as an application where users can work with different models, such as Gema 4, and interact with them using natural language, with the model loading into memory and utilizing the GPU or CPU to respond to user input, 2m6s.
• Olama is intelligent enough to abstract users from the problem of configuring drivers for CUDA on Nvidia or AMD, making it a convenient tool, and it is widely used, including by the user themselves, who has created a Windows application to monitor Olama's activity and provide updates on which models are being loaded 10s.

Lama Monitor Application and Its Functionality

• The user has created a small Windows application, called Lama Monitor, which runs in the background and displays a popup with information about the models being loaded by Olama, helping to keep track of the execution of automated tests 2m6s.
• Olama is not open source, but it is believed to be programmed in C, and it may have additional layers or libraries, such as Lama CPP, which was previously used, although it is unclear if it has been migrated to other languages like Go or Rust 4m30s.

Team Discussion and Connection to Microsoft Build

• The user mentions that they have a team of professionals, including Francis and Cristian, and they discuss the topic of Olama and its potential connection to the Cali de Build, which has over 300 published sessions, with around 250 available to view online, along with downloadable materials 8m40s.
• The user's Windows application is designed to work with Olama and provide real-time updates on the models being loaded, which is particularly useful when running automated tests, and it is not clear if Olama works with other programming languages like Go 6m20s.

Using Build CLI to Filter Sessions

• The process of selecting what to view can be challenging due to the numerous options available, but utilizing a project like Olama Monitor can help streamline this process by allowing access to Copilot, a tool that facilitates the installation of plugins such as Microsoft Build 10s.
• Installing a plugin like Microsoft Build involves a straightforward process where the plugin is downloaded and installed, after which the application needs to be restarted to complete the installation, and this process is considered the basics of installing a plugin 42s.
• Once the Microsoft Build plugin is installed, it can analyze the repository and code of a project, identifying the context, dependencies, and purpose of the project, such as determining that it is a Windows application created with Windows Presentation Foundation in DNET 10 that utilizes Olama 2m6s.
• The plugin can then suggest relevant sessions related to the project, making it easier to find useful information among the numerous available sessions, which can number in the hundreds, and this feature is considered fascinating 2m6s.
• The analysis performed by the plugin provides a comprehensive overview of the project, including its integration with other tools and systems, such as OLAMA and the SIS Tray, and helps identify areas of interest and potential improvements 2m6s.
• The BRK 260 session is about building applications with local technology, and it is possible to use Foundry Local, a Foundry technology that allows executing local models on Windows, Linux, and Mac, instead of Olama, to achieve the same results 10s.
• Foundry Local is a technology that enables running models locally on various operating systems, although it is unclear if Linux is supported, and it can be used as an alternative to Olama for building applications 42s.
• The session BRK 260 can be found by searching for it in the build section, and it is scheduled to take place on June 3rd in San Francisco, online, and will be recorded, providing an opportunity to learn about building applications with local technology 1m30s.
• The Build CLI can be used to filter relevant sessions based on a specific project, and it provides a list of sessions, including table talks, which are conversations with moderators that are not recorded, allowing users to focus on the most relevant content for their projects 2m6s.
• The Build CLI can help filter sessions based on a specific project, and it can provide a list of online sessions that are relevant to the project, making it easier to find the most useful content 3m30s.
• The ability to filter sessions using the Build CLI is useful for projects, as it helps to narrow down the content to the most relevant sessions, and it can be used to ask questions about the sessions, such as which ones are online and relevant to a specific project 4m20s.

Session BRK 260 and Foundry Local

• The BRK 260 session is about building applications with local technology, and it is possible to use Foundry Local, a Foundry technology that allows executing local models on Windows, Linux, and Mac, instead of Olama, to achieve the same results 10s.
• Foundry Local is a technology that enables running models locally on various operating systems, although it is unclear if Linux is supported, and it can be used as an alternative to Olama for building applications 42s.
• The session BRK 260 can be found by searching for it in the build section, and it is scheduled to take place on June 3rd in San Francisco, online, and will be recorded, providing an opportunity to learn about building applications with local technology 1m30s.

Repository and Session Filtering

• The integration of certain features is quite interesting and could be useful for investigating how things work behind the scenes, such as tokenization and contextualization of people, and making recommendations based on the project in a connected repository 10s.
• A repository can be local or on Git, and it's interesting to see how things are managed, with the possibility of including it in a CL to realize certain goals, and a task will be created to include it 42s.
• There is a publicly available, open-source repository with a URL that can be shared, containing 80% TypeScript and 16% JavaScript, which makes API calls to obtain sessions and is a useful template for applications or event sites 2m6s.
• The repository is simple, practical, and useful, providing value and functionality, and it can be used as a template for other applications, with the ability to view and modify it since it's open and public 2m6s.

Relevant Sessions for Olama Monitor

• The sessions online are relevant for the Olama monitor, with specific codes such as OD SP929 and OD SP902, and it's possible to search for them on the site to find more information, with OD possibly standing for "on demand" 4m30s.

Introduction to 'uno' Platform

• There is a platform called "uno" that provides a library for creating applications on various platforms, including Windows, Linux, Android, and Mac, and it gives modern applications with responses in Spanish 6m0s.
• A link was previously shared, but it's not remembered, and Francis is asked to provide the link again, with the possibility of prioritizing certain topics, such as sessions, to make it easier to find the necessary information 8m0s.

Exploration of AI Models and Tools

• The discussion involves exploring various models and applications, including Olama, Foundry Local, and LM Studio, which are all free and can be installed locally to run models 10s.
• Olama is an application that allows users to run different models, including Quen, which comes in various versions, and Coen 3.6, a large model that requires significant hardware capabilities 2m6s.
• Some models, such as those with 27 billion parameters, can be quite large, weighing around 30 GB or more, and may not be executable on local hardware, while others like MLX are specifically designed for Apple devices 2m6s.

Olama Monitor Installation and Functionality

• The Olama Monitor application can be installed using specific lines of code, and it allows users to monitor the performance of their models, including CPU, memory, and GPU usage 4m30s.
• When running models like Lama 32 and Gema 4, the application displays the model's performance and resource usage, and it can also load multiple models simultaneously, demonstrating its capabilities 6m40s.
• The application's functionality is attributed to its underlying architecture, which enables it to manage multiple models and allocate resources efficiently 8m50s.

Commands and Features of Olama

• Olama provides commands such as Olama PS to show all working processes and Olama LS to display all downloaded models, and the application is designed to monitor and execute these commands at regular intervals to provide updates on loaded and unloaded models 10s.
• A pull request from Pablito, a colleague, has been made to allow model downloads, and there is also a command called Olama Pur that can be used to free up memory when needed 1m5s.

Notifications and Integration with Copilot

• The goal is to have the application work with Windows notifications, so when Olama is working, a popup notification can appear to inform the user that a model has been loaded into memory 2m6s.
• The application will be tested using Copilot, with the task of showing a Windows notification when Olama changes state, is activated or deactivated, or when a model is loaded or unloaded 3m15s.
• Copilot is being used with the automatic model selection, which chooses the most suitable and cost-effective model for the task, in this case, the cloud-based Haiko model 4m20s.
• The pros of the idea, as evaluated by Copilot, include the need for the user to see the system tray, native and consistent functionality, and usefulness for presentations, with the option to control and customize notifications 5m30s.
• The trade has variability between Windows 10 and 11, and it is assumed that Windows 11 is being used, with no multi-platform testing available because Windows does not capture quick changes 10s.

Implementation Details and Testing

• The implementation of a feature is being discussed, including its functionalities and settings, such as the type of events to be shown as Windows notifications, and the use of Copilot 42s.

Microsoft Student Ambassador Program Overview

• The conversation turns to the topic of the Microsoft Student Ambassador program, which provides benefits to students, including access to Copilot, and the requirements to join the program, such as being over 18 years old and enrolled in a higher education institution 2m6s.
• To apply for the Microsoft Student Ambassador program, students must register and join the program's Discord community, with no need for a video presentation, and can then progress through the program's levels, which have been updated and simplified 4m30s.
• The program's levels, previously known as Alfa, IT beta, and Gold, have been renamed, with the first level now being called Microsoft Student Ambassador, and the program offers incentives, such as support for organizing events, including funding for catering and snacks 6m20s.
• The program is led by directors, including Euri and Pablo Vera Mendi, who provide guidance and support to the ambassadors, and the program has undergone changes over time, with the goal of making it more accessible and beneficial to students 8m10s.

Benefits and Organization of Events

• The benefits of organizing events include receiving items such as cool swags, including poleras, mochilitas, stickers, and light bars for decoration, which can be given away or raffled off 10s.
• The Microsoft Student Ambassadors program can be taken advantage of by students who are enrolled in a university, and it offers various benefits, including the chance to register for the program and receive rewards 42s.
• The Microsoft Student Ambassadors program has undergone changes, previously being known as Microsoft Learn Student Ambassadors, and now offers a wider range of technologies and opportunities for students to learn and grow 1m6s.

Program Requirements and Eligibility

• The program has different levels, including senior, and offers benefits such as licenses, My Visual Benefit, and the opportunity to claim rewards, which can be exciting for students 1m30s.
• In some countries, including Bolivia, the Microsoft Student Ambassadors program is not well-known, but organizing events and giving talks can help raise awareness and encourage students to participate 2m6s.
• To be eligible for the program, students typically need to be enrolled in an institution and have a form of identification to demonstrate their student status, which can be beneficial for those who want to take advantage of the program's benefits 2m42s.

Updates and Registration Process

• The program is constantly being updated, and students can stay informed about changes and new opportunities through various channels, including GitHub and videos 3m0s.
• The analysis of a case takes approximately 10 minutes, during which the student ID is verified, and after acceptance, resources are made available within a certain number of days, allowing the implementation of various tools such as Copilot in Visual Code 10s.
• The GitHub Student Developer Packs program offers numerous benefits, including access to various tools and resources, and it is recommended to apply for the program when new changes are implemented, as there are many pros to being part of it 42s.

GitHub Student Developer Pack

• GitHub has several programs available, not just the Student Developer Pack, and it is possible to sign up for these programs through the education.github.com/pac website, where students and professors can register by providing their school name, email, and other basic information 2m6s.
• The program provides access to tools like Copilot, which can be extremely helpful in amplifying productivity, and it is essential to learn how to utilize these tools effectively to get the most out of them 2m6s.

David Fowler and Aspire Project

• David Fowler, a distinguished engineer at Microsoft with over 15 years of experience, has written an article about how his team is working with Aspire, which is recommended reading to understand how to work with GitHub tools in the complete development cycle of a real production product 4m30s.
• It is advised to read David Fowler's article to gain more context about working with GitHub tools and to learn from his experience as the cocreator of several relevant technologies, including ASPNET core and SignalR 6m15s.
• The team has been transparent about their use of artificial intelligence in the development cycle to generate documentation, perform tests, and create code, sharing what works and what does not, as well as their plans for future testing 10s.
• David, who is from Barbados, the same country where Rihanna was born, has been sharing his experiences and knowledge, including through streams where the team showcases their work on Aspire, a tool for orchestration in .NET, JavaScript, and TypeScript 42s.
• The team has been sharing their experiences with artificial intelligence on Twitter and LinkedIn over the past six months, providing a valuable resource for those interested in learning more about the topic 2m6s.

AI in Development and Productivity

• Artificial intelligence can be used to automate boring tasks, allowing programmers to focus on more interesting and challenging aspects of their work, such as those that require creativity and problem-solving skills 2m6s.
• The use of artificial intelligence can also help with tasks that are often neglected or put off, such as those found in a project's backlog, and can be a useful tool for programmers 2m6s.
• David is described as a knowledgeable and kind person, who is willing to share his expertise and experiences with others, and is an example of how experts in the field can be approachable and helpful 4m30s.

Account Management and EI Discussion

• There is a discussion about the creation of accounts and how they can be transferred between plans, but the details are not fully understood and would require further clarification 6m10s.
• The conversation touches on the topic of making things boring with EI, which is a natural approach, and it is not wrong to do so, as many people, including Cristian, Francis, David Fler, and others, are working on it 10s.

Windows Notifications and Application Testing

• There is a mention of checking the monitor to see if the process has finished and if Windows notifications are working, and also a greeting to Luis, who is commenting and is a friend from Mexico but currently living in Eastern Europe 42s.
• The discussion moves on to the code, where a review is being done, and it is noted that the process has finished, and files have been created, including a notification service that sends Windows notifications, with settings to enable and disable it 2m6s.
• The settings also include notification events, bounce, and a form that is the Windows view, with a front and view model to work with, and it has three sections: Olama, model, and resources, with additional features such as CPU alerts and state tracking 2m6s.
• The application is launched, and it is noted that the settings option is missing from the menu, which is a problem, and a request is made to fix it, with a humorous comment about being polite to AI systems in case they conquer the world 4m30s.

Uninstallation and Virtual Machine Setup

• The decision is made to uninstall Olama Monitor to avoid complications with the local version being executed, and it is mentioned that this process should be done on a separate virtual machine to avoid issues 6m40s.
• The process of working with various settings and tools is underway, and errors are expected to occur, which will require adjustments and changes to be made 10s.

Settings and UI Adjustments

• A settings menu is being modified, and items are being added to it, including a service that was previously created, and the goal is to make the settings work properly 42s.
• The settings window is currently disorganized, and there are many settings available, including a model that can be loaded and unloaded, and the plan is to enable all settings and test them 2m6s.
• A button to record settings is missing, and it is assumed that the settings will be saved automatically, but this is not the case, and a save button needs to be added 2m6s.
• Olama, a tool, is being used, and its status is being checked, and when it is closed, a notification appears indicating that it is offline, and when it is restarted, it should appear again 4m10s.
• The process of working with virtual machines and automated tests is complex, and notifications are needed to indicate when certain actions have been completed, such as when a model has been loaded into memory 6m20s.
• The current settings and notifications are not functioning as expected, and changes are needed to make them work properly, including adding a save button and ensuring that settings are saved automatically when the settings form is closed 8m30s.

Font and Model Adjustments

• The discussion involves working with a model to improve font visibility in a certain resolution, and the model needs to understand the requirements and make adjustments, which may take some time to compile and fix, 10s.
• There are notifications from Windows, and a list of Build sessions to attend has been suggested, with 10 minutes left to work on the current task and potentially explore other topics, 42s.

Programming and Library Discussion

• The conversation turns to programming and libraries, with a request to share experiences or libraries used in daily work, and Francis is asked to share his recent work, which involves the MCPSDK protocol, 2m6s.
• Francis mentions working with the MCPSDK protocol, Microsoft Antis Kernel, and Semantic Kernel, and also exploring image and audio generation using MCP, 2m6s.
• The topic of migrating to Microsoft AN framework is mentioned, and Cristian's video on the subject is shared, with a plan to discuss video and audio generation using new models from Foundry and Open AI after the current task is completed, 2m6s.

Foundry and Open AI Models

• A demo on Foundry is mentioned, with a link to a YouTube video in the comments, which shows how to create a server and generate interesting applications and agents using MCP, 2m6s.
• New models have been released by Open in recent weeks, which are updated versions of their real-time models, allowing for real-time conversations and translations, including a feature to translate audio from one language to another, such as from Spanish to English or Japanese, and these models are considered to be very good 10s.
• These models can be used in various applications, including Foundry, where users can create agents, connect them to tools like web search, knowledge bases, and security features, and activate a voice mode to test the models 1m30s.
• The voice mode allows users to interact with the models using voice commands, and it can be used to retrieve information on various topics, such as the current weather in a specific location, like Toronto 2m6s.
• To use the voice mode, users need to share their computer audio, and it may require sharing the entire screen or window to work properly, and once set up, the model can respond to voice commands and provide information on the requested topic 4m0s.

Voice Mode and Real-Time Interaction

• The models can provide detailed information, such as the current temperature, weather conditions, and time of day, for a specific location, and they can also ask for clarification or confirmation before providing the information 6m0s.
• The test of the voice mode involved asking the model to provide information on the current weather in Toronto, and the model was able to respond with the current temperature, weather conditions, and time of day, demonstrating its ability to provide accurate and up-to-date information 8m0s.

Tool Interaction and Language Settings

• The conversation starts with a discussion about giving instructions to a tool that can search online, and it is decided that the tool should respond in Spanish and speak like a pirate 10s.
• A question is asked about the climate, and the tool is asked to provide the current temperature in Celsius in Cochabamba, Bolivia, but it starts providing a URL in English instead of the temperature 42s.
• The tool's response is not what was expected, and it is decided not to force the issue, and instead, the conversation moves on to discussing the code used to interact with the tool, which can be written in languages such as Python, TypeScript, JavaScript, or C# 2m6s.

Fundry and Medical Diagnosis Project

• The conversation also touches on the topic of Fundry, which is a platform that can be used with ambassadors, and it has features such as credits, evaluation, monitoring, and a playground for testing 2m6s.
• There is a mention of a previous project that involved using machine learning for medical diagnosis, including voice analysis and natural language processing, which was presented and found to be interesting 4m10s.
• The project used APIs and scripts to provide responses and prediagnoses for patients, and it is suggested that this topic should be discussed further because it may be of interest to many people 6m0s.

Improvements and Third-Party Tools

• The discussion involves various topics, including Bill, Foundry, and improvements made to the Olama Monitor, which is not yet fully improved but is almost there 10s.
• André asks if it is possible to view conversations in a more user-friendly way, and the response is that there are different views available in Foundry, as well as third-party tools that can abstract the details and present them in a more human-readable format 42s.
• The third-party tools that can help view conversations in a more user-friendly way are not named, but it is suggested that interested parties can contact the speaker on LinkedIn to learn more about these tools 1m6s.

Closing Remarks and Upcoming Events

• The conversation comes to a close with expressions of gratitude and announcements about upcoming events, including the return of Andrea as the host the following week 2m6s.
• The discussion ends with farewells and greetings to friends, including Francis and Cristian, and a promise to meet again in the future 2m6s.