Skip to Content
View all events

Deploying NLP Models in Production using MLOps

Published by Pearson

Beginner to intermediate content levelBeginner to intermediate

Building Transformer Models and Handling Model Drift

This training provides an overview to the end-to-end Natural Language Processing pipeline including the initial model training, production deployment and serving, model evaluation, and continuous training cycles to combat model/data drift.

We look at various tools including PyTorch serve and MLflow to manage model versions and deploy them in a production infrastructure. We also see several code examples throughout the training around a semantic search use-case using BERT to help solidify the theoretical concepts being introduced.

What you’ll learn and how you can apply it

By the end of the live online course, you’ll understand:

  • The end-to-end Natural Language Processing pipeline
  • How to train the initial model, model evaluation, and how to handle model/data drift
  • Manage model versions and deploy using PyTorch and MLflow

And you’ll be able to:

  • Productionize modern transformer-based NLP models
  • Understand the parts of an MLOps infrastructure and pipelines for deploying models
  • Detect and handle model drift

This live event is for you because...

  • You’re an advanced Machine Learning Engineer with experience with ML, Neural Networks, and NLP
  • You’re interested in state-of-the art NLP Architecture
  • You’re interested in productionizing NLP models
  • You are comfortable using libraries like Tensorflow or PyTorch

Prerequisites

  • Python 3 proficiency with some familiarity working in interactive Python environments including Notebooks (Jupyter / Google Colab / Kaggle Kernels).
  • Comfort using libraries like Tensorflow or PyTorch

Course Set-up

  • A GitHub repository with the slides / code / links will be provided
  • Attendees will need to have access to the Notebooks in the GitHub

Recommended Preparation

Recommended Follow-up

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Segment 1: Introduction to Transformers and semantic search Length (30 min)

  • Introduction to Transformers and how they are used to process text
  • Introduction to Transfer Learning and modern NLP tasks
  • Overview of our semantic search use-case using BERT

Segment 2: Serving Transformer models in production Length (40 min)

  • Introduction to model versioning with mlflow
  • Introduction to model serving with pytorch serve
  • Using mlflow and pytorch serve to productionize NLP models

Break: 10 min

Q&A: 10 min

Segment 3: Detecting Model Drift Length (30 min)

  • Learning the four types of model and data drift
  • Calculating model and data drift through statistical tests
  • Identifying the causes of model drift

Segment 4: Handling Model Drift Length (40 min)

  • Detecting and removing bad training data from corpora
  • Taking advantage of online and batch learning to update NLP models

Break: 10 min

Q&A: 10 min

Segment 5: The End-to-End NLP Pipeline Length (40 min)

  • Optimizing ML infrastructure for scalability
  • Integration of Feature Stores across an ML organization
  • Introduction to MLOps and CI/CD for ML/NLP

Segment 6: Course wrap-up and next steps Length (20 min)

  • Next steps / further resources
  • Final Q/A

Your Instructor

  • Sinan Ozdemir

    Sinan Ozdemir is the founder of Crucible, an AI factory platform that helps teams convert existing workflows into custom models. He is a Y Combinator alum, AI & LLM Advisor at Tola Capital, and the author of multiple books on data science and machine learning including Building Agentic AI, Quick Start Guide to LLMs, and Principles of Data Science. Sinan is a former lecturer of data science at Johns Hopkins University and the founder of Kylie.ai, an enterprise-grade conversational AI platform (acquired 2014). He holds a master's degree in pure mathematics from Johns Hopkins University and is based in San Francisco, California.

    linkedinXlinksearch

Skill covered

MLOps