Microsoft wants developers to evolve into AI developers. It’s a predictable enough progression i.e. first, developers were mainframe developers, then they became PC developers, mobile developers and then cloud developers. Next, logically, software engineers will have to become automation-centric AI developers with closer proximity to the models and engines that enable our new intelligence services.
To facilitate this shift, Redmond is of course championing the use of Microsoft Azure AI services as a route to building custom AI applications. To achieve the critical mass it so dearly seeks, Microsoft is partnering with GitHub to enable its 100 million+ developers to build AI applications directly on the GitHub platform.
Although Microsoft acquired GitHub in 2018 and the platform has subsequently been bolstered by CI/CD pipeline tools and wider security capabilities, it remains a comparatively independent entity in the style of a Red Hat inside IBM.
GitHub Models Marketplace
Microsoft has now made Azure AI Studio model selection functions available to developers through GitHub Models, a “marketplace” portal where software engineers who want to develop a generative AI application can find and experiment with AI models for free before considering token-based payment once an application moves to production. Efforts from Microsoft here also include the provision of APIs designed to enable production-ready AI applications.
GitHub Marketplace itself exists as a “free playground” where programmers can adjust model parameters and submit prompts to see how an AI model responds. This work also features integrations with Codespaces and Microsoft Visual Studio Code. GitHub Codespaces provides a secure development environment with a selection of prebuilt built-in resources that enjoy native integration with the GitHub platform.
GitHub Models came about in August of this year as a means of enabling the rise of the AI engineer with access to leading large and small language models.
The Rise Of The AI Engineer
“As AI model innovation accelerates, Azure remains committed to delivering the leading model selection and greatest model diversity to meet the unique cost, latency, design and safety needs of AI developers. Today, we offer the largest and most complete model library in the market, including the latest models from OpenAI, Meta, Mistral and Cohere and updates to our own Phi-3 family of small language models,” notes Asha Sharma, corporate vice president of AI platform, Microsoft.
With GitHub Models, Sharma suggests developers can explore the latest models along with other AI innovations such as next-generation “frontier models” i.e. those models that rank as large scale with many parameters and complex algorithmic logic such as Google Gemini, Anthropic’s Claude, AlphaFold’s DeepMind and indeed GPT-4 from OpenAI.
“For most of us, learning to be a developer didn’t happen on a linear path in the classroom. It took [plenty of] practicing, playing around and learning through experimentation. The same is true today for AI models. In the new interactive model playground, students, hobbyists and startups can explore the most popular private and open models from Meta, Mistral, Azure OpenAI Service Microsoft and others with just a few clicks and keystrokes,” blogged GitHub CEO Thomas Dohmke.
Many Modes & Modals Of Models
This entire dovetailing operation has likely come about as a result of the sheer size and scope of AI model development today. Twinned with the need to straddle deployment attack surfaces spanning cloud, edge, general-purpose and perhaps more task- or industry-specific application and data service deployments, the complexity factor out there is dizzying. GitHub Models aims to open the door for developers to experiment with multiple models, simplifying model experimentation and selection across the Azure AI catalog so that engineers can compare models, parameters, and prompts.
By making Azure AI an open, modular platform, Microsoft’s Sharma says the company aims to help our customers rapidly go from idea to code to cloud. With Azure AI on GitHub, developers can use Codespaces to set up a prototype or use the Prompty extension to generate code with GitHub Models directly in Microsoft Visual Studio Code.
“Increased model selection gives developers the broadest range of options for the individual applications they are building. But each model naturally brings with it increased complexity. To counteract this, we’re making it incredibly easy for every developer to experiment with a range of models through the Azure AI model inference API. Using this single API, GitHub developers can now access a common set of capabilities to compare performance across a diverse set of foundational models in a uniform and consistent way, easily switching between models to compare performance without changing the underlying code,” said Sharma.
On the road ahead, Microsoft says it will expand its integration even further. Surprisingly specific about what this means, this is Redmond’s commitment to bringing Azure AI’s language, vision and multi-modal services to GitHub, along with additional Azure AI toolchain elements. Organizations with an existing Azure subscription can purchase GitHub products via self-service, directly through Microsoft sales or via third-party solution providers and can adjust the number of GitHub seats as needed.
