Syllabus.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MLOps: A Gentle Introduction to Machine Learning Deployment\n",
    "\n",
    "\n",
    "## 1. Summary\n",
    "\n",
    "- Term, CRN and Number of Credits: Spring 2024,  44420 (IDS 594) and 2 resp.\n",
    "- Time: Wed 3.00 - 5.30 PM\n",
    "- Location: SH 320\n",
    "- Staff:\n",
    "     - Instructor: [Dr. Theja Tulabandhula](https://theja.org)\n",
    "       - Office Hours: Wed 1.30 - 2.30 PM at UH 2404/zoom by email appointment\n",
    "       \n",
    "## 2. Objective\n",
    "\n",
    "This practice-oriented course surveys modern best practices around getting machine learning (ML) models into production. It complements IDS 572, IDS 575 and IDS 576, which is to learn multiple ways of operationalizing machine learning workflows and models in the context of the larger business end-goals. The course is also complementary to IDS 561. We will gain a better understanding of strategies for model management, monitoring and deployment. Further, we intertwine these topics with online experimentation techniques (A/B testing) and software engineering ideas such as version control, containerization, and continuous integration/continuous deployment.\n",
    "\n",
    "A tentative list of topics is as follows:\n",
    "\n",
    "- Deploying ML models using web servers (e.g., using Flask)\n",
    "- Containers for machine learning: the Docker ecosystem and Kubernetes\n",
    "- Model management: model tracking logging, scheduling of training, A/B testing\n",
    "- Distributed and streaming setups: Apache Spark and Kafka\n",
    "\n",
    "## 3. Course Logistics\n",
    "\n",
    "### 3.1. Dates\n",
    "\n",
    " - 01/10: lecture\n",
    " - 01/17: lecture\n",
    " - 01/24: lecture\n",
    " - 01/31: lecture\n",
    " - 02/07: lecture\n",
    " - 02/14: lecture\n",
    " - 02/20: project final report+code due\n",
    " - 02/21: lecture slot for live student project presentations I\n",
    " - 02/28: lecture slot for live student project presentations II"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "### 3.2. Topics\n",
    "\n",
    "\n",
    "- Chapter 1: Serving ML Models Using Web Servers\n",
    "  - Learning Goals:\n",
    "    - Be able to set up a Python environment\n",
    "    - Be able to set up a jupyter session with SSH tunneling\n",
    "    - Be able to secure a web server\n",
    "    - Be able to use Flask to serve a ML model\n",
    "\n",
    "- Chapter 2: Serving ML Models Using Serverless Infrastructure\n",
    "  - Learning Goals:\n",
    "    - Be able to differentiate hosted vs managed solutions\n",
    "    - Assess devops effort for web server vs serverless deployments\n",
    "    - Be able to deploy a ML model using Google Cloud Functions and AWS Lambda Functions\n",
    "\n",
    "\n",
    "- Chapter 3: Serving ML Models Using Docker\n",
    "  - Learning Goals:\n",
    "    - Be able to reason the pros and cons of container technologies\n",
    "    - Be able to differentiate containers from virtual machines\n",
    "    - Be able to create a new Docker image using Dockerfile\n",
    "    - Be able to upload the image to a remote registry\n",
    "\n",
    "- Chapter 4: Kubernetes for Orchestrating ML Deployments\n",
    "  - Learning Goals:\n",
    "    - Understand the uses of Kubernetes\n",
    "    - Be able to set up a single node Kubernetes cluster using `kubectl` and `minicube`\n",
    "    - Be able to serve a prediction model on a container in the Kubernetes cluster\n",
    "    - Be able to deploy a prediction model on Google Kubernetes Engine (GKE)\n",
    "\n",
    "- Chapter 5: ML Model Pipelines\n",
    "  - Learning Goals:\n",
    "    - Learn how to manage a model building workflow\n",
    "    - Learn how to set up automated jobs using `cron`\n",
    "    - Learn the basics of Apache Airflow\n",
    "    - Learn a managed workflow tool (Google Cloud Composer)\n",
    "\n",
    "- Chapter 6: PySpark Ecosystem\n",
    "  - Learning Goals:\n",
    "    - Understand the components on a spark cluster\n",
    "    - Be able to use PySpark and spark dataframes\n",
    "    - Be able to use models from MLLib\n",
    "    - Be able to work with a managed solution such as Databricks\n",
    "\n",
    "- Chapter 7: Streaming Model Deployments\n",
    "  - Learning Goals:\n",
    "    - Understand the difference between a streaming model deployment workflow vs a batch model deployment workflow\n",
    "    - Learn the basics of streaming with Apache Kafka\n",
    "    - Be able to differentiate between a batch Pyspark workflow and a  Pyspark streaming workflow\n",
    "\n",
    "\n",
    "- Chapter 8: Online Experimentation\n",
    "  - Optional Reference: https://docs.aws.amazon.com/sagemaker/latest/dg/model-ab-testing.html\n",
    "  - Learning Goals:\n",
    "    - Know the considerations for A/B testing of models before full rollouts\n",
    "    - Be acquainted with a few statistical hypothesis tests and how sample sizes are determined\n",
    "    - Be able to create simple experiments using `planout` and a flask based deployment setup"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Course Materials\n",
    "\n",
    "- A [slide deck](https://docs.google.com/presentation/d/1IxHVGTQu4HIyDPDPPL9azAH-jeQcLylm4j7af1CZbm8/edit?usp=sharing) covering the tools and ideas for each chapter.\n",
    "- Example implementations are available at [https://github.com/thejat/mlops-code-examples](https://github.com/thejat/mlops-code-examples). If you see errors, please raise a issue or make a PR!\n",
    "\n",
    "\n",
    "## 5. Project\n",
    "\n",
    " - The objective will be to demonstrate a deployment of an existing machine learning model that you have access to. A suitable documentation of this process along with the complete set of scripts/codes/commands used is to be submitted on the due date (with no exceptions). See the [project page](Project.ipynb) for the due date and more detailed instructions.\n",
    " - You should form groups of _strictly_ $2$ students each for the project component and update your details in the spreadsheet on [Blackboard](https://uic.blackboard.com). Reach out to your classmates early using the same spreadsheet. Because this is a group project, a commensurate effort is expected, and each member's contributions needs to be explicitly reported in the final submission.\n",
    "\n",
    " \n",
    "## 6. Grading\n",
    "\n",
    " - There are no graded assignments or exams for this course.\n",
    " - Grades will be assigned based on:\n",
    "    - the project (90\\% equially split between the criteria defined in the project page), and \n",
    "    - course participation *strictly necessary* (10\\% that includes but is not limited to attendance, interaction with the instructor and the TA, and how well you support your group)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. Miscellaneous Information\n",
    "\n",
    " - This is a 2 credit graduate level course offered by the Information and Decision Sciences department at UIC.\n",
    " - See the [academic calendar](http://catalog.uic.edu/ucat/academic-calendar/) for the semester timeline.\n",
    " - Students who wish to observe their religious holidays (http://oae.uic.edu/religious-calendar/) should notify the instructor within one week of the first lecture date. \n",
    " - Contact the instructor at the earliest, if you require any accommodations for access to and/or participation in this course (see additional details below).\n",
    " - Refer to the academic integrity guidelines set by the university.\n",
    "\n",
    "#### Special Accommodations\n",
    "\n",
    "UIC has the [Disability Resource Center (DRC)](https://drc.uic.edu/) to help students needing special accommodations. If you have any condition, such as a physical or learning disability, which will make it difficult for you to carry out the work as outlined in the syllabus or which will require academic accommodations, you can go to the DRC for assistance, and should do so at the beginning of the semester or as early as possible. \n",
    "\n",
    "If a student needs remote learning accommodation due to a physical condition, such as if they are immuno-compromised and especially vulnerable to Covid, then the student can submit a formal request for remote learning accommodation through the DRC. This is needed in order to maintain fairness and consistency in our attendance policy."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}