Moving machine learning models from experimentation to production remains one of the biggest challenges in data science teams today. In this session, we'll explore ML Accelerator, an end-to-end MLOps platform that transforms your ML workflow from training models to experiment tracking using an automated infrastructure. Through architecture walkthroughs and examples, you'll discover how we leveraged GitHub Actions, MLFlow, and containerised deployments on AWS to create a data scientist friendly platform that handles experiment tracking, model registry, and seamless production deployments. Whether you're struggling with model versioning, experiment reproducibility, or production deployment complexity, this talk will provide practical insights and architectural approach you can adapt to accelerate your own ML operations. You'll learn how thoughtful platform engineering can streamline ML deployment workflows while addressing key concerns around ease of training and tracking experiments.

Natural Language Processing (NLP) solutions have become an important part of businesses as they scale handling customer issues and queries. It offers better customer engagement and provides relevant information on finger tips. One of the major challenges is being able to communicate across various channels and languages quickly, effectively and securely.

In this talk, we will present our journey on how we built a bespoke event driven NLP communication platform which is generic enough to be used across multiple channels and languages and is capable to switch between NLP engine and human agent in an ongoing conversation to provide more relevant information . We will talk about challenges with training NLP models for handling multiple languages and classes of queries. We will also look at how we built an SDK chat component that can be integrated with any mobile application.