Have you ever deployed a machine learning project to production with the same principles as a software project? I did - I failed. But, on the way, I learned many essential factors to run ML in production environments successfully! So there is more to it than just deploying a data scientist Jupyter notebook to AWS. This talk will go through some common pitfalls of running machine learning in production settings. We will start with the requirements and work through the data acquisition and model-building phase. We explore beyond the current MLOps hype and try to understand what it takes to run a successful project that is ready to ripe like a fine wine rather than old milk.
Have you ever deployed a machine learning project to production with the same principles as a software project? I did - I failed. But, on the way, I learned many essential factors to run ML in production environments successfully! So there is more to it than just deploying a data scientist Jupyter notebook to AWS. This talk will go through some common pitfalls of running machine learning in production settings. We will start with the requirements and work through the data acquisition and model-building phase. We explore beyond the current MLOps hype and try to understand what it takes to run a successful project that is ready to ripe like a fine wine rather than old milk.