IML4E Meetup series started in Budapest
Can we produce intelligent software the way the automotive industry produces vehicles? This question was posed by Nayur Khan, one of the speakers at the kick-off event of the lecture series initiated by the #IML4E project on the industrial application of machine learning and the potential of MLOps. A total of 3 speakers and over 60 guests in attendance engaged in a lively discussion about the opportunities and challenges of implementing MLOps in companies. Jürgen Großmann, the coordinator of the #IML4E project, reported in his presentation "An introductory talk to the world of MLOps" about the basics of #MLOps and emphasized the enormous opportunities to make industrial processes more effective by using data and ML. At the same time, he emphasized that currently the majority of planned ML projects do not achieve the expected goal. A large part fails even before deployment. In his talk he presented solutions for industrialized ML and innovations from the IML4E project with which the introduction of #MLOps can succeed.
Nayur Khan shared best practices concerning AI, data, diversity, and software engineering. In his talk "Scaling AI like a tech native" he shared his experiences in helping organizations move away from pilots and experiments with AI to industrialized implementations that run reliably at scale. Nayur is a partner with McKinsey Digital's London office and also part of the QuantumBlack, AI by McKinsey team. He is focused on helping organizations build capabilities to industrialize and scale artificial intelligence to improve performance.
In the last presentation, "Practices and Infrastructures for MLOps," Dennis Muiruri, Researcher at University of Helsinki, explained established artificial intelligence engineering approaches. Based on a series of interviews with 16 organizations across various domains in Finland, he mapped the current situation in the use of MLOps tools and was able to conclude that there is a great need for research and development especially in the area of testing and monitoring of ML-enabled systems.
In and through their presentations, all three speakers confirmed the need to design MLOps tools and processes in such a way that even organizations that have had little exposure to ML can gain the ability to make sense of their data. With the IML4E framework, #IML4E can provide direction in this area.