TradeRev Is a Machine Learning Vehicle Appraisal / Auctioning System

Copied URL with current time.
0:00 / 0:00
avatars/amit-jain.jpg
Amit Jain
https://www.traderev.com/en-us/

In this episode of Running in Production, Amit Jain goes over building an auctioning system that uses machine / deep learning and is powered by Flask and Python. It’s all hosted on AWS and has been up and running since mid 2011.

Amit goes over a few machine learning libraries, refactoring a 100k+ line monolith into microservices without any automated tests, the importance of machine learning accuracy, using a bunch of AWS services to deploy a large site, treating your infrastructure as code and more.

Topics Include

  • 3:58 – Amit lead a team of ~10 R&D engineers responsible for Data Science / ML
  • 4:33 – Roughly 1,000 cars a day are being traded with 8-10k auctions / bids per day
  • 5:15 – Motivation for using Flask and Python
  • 6:55 – Scikit-Learn and TensorFlow for machine / deep learning
  • 7:39 – Did things start off with multiple microservices or was it a monolith early on?
  • 9:41 – There’s about 80,000 to 120,000 lines of code across 200-300+ Python files
  • 10:14 – The huge refactor to microservices was done without automated tests initially
  • 11:11 – After the refactor now there’s 86% test coverage which is enough to be confident
  • 12:24 – Flask-Restplus is the main library used to build their RESTful APIs
  • 12:43 – Other notable libraries were gunicorn and boto3 (AWS SDK for Python)
  • 13:05 – Locust is an open source load / performance testing tool
  • 13:40 – With machine learning, speed is important but accuracy is even more important
  • 15:30 – gunicorn is very compact, performant and easy to configure
  • 16:28 – Most caches were in memory and they used Amazon DynamoDB
  • 17:09 – The primary database is MySQL running on Amazon RDS
  • 18:04 – SQLAlchemy is used on the Python side as an ORM
  • 19:29 – Docker is sort of being used in development
  • 21:02 – The platform runs on AWS with Lambda, API Gateway and AWS Fargate with ECS
  • 22:24 – What is AWS Fargate and what does it allow you to do?
  • 23:48 – Scaling with Fargate while using auto scaling policies and configuration
  • 26:28 – Taking advantage of the cloud and setting up load balancing with configuration
  • 28:04 – How do you deal with secrets when using Fargate / ECS?
  • 30:02 – What about logging and metrics? Are you exclusively using all of AWS’ services?
  • 31:12 – What about error reporting, such as getting notified if an error happens
  • 31:34 – The deploy process from development to production (includes CI / CD with Jenkins)
  • 33:26 – A Walk through of how the different AWS services come together
  • 36:54 – Terraform is being used to manage the infrastructure as code (valuable tool)
  • 40:04 – Database backups were performed by the DevOps team
  • 40:41 – Best tips? Start slow and expect failures, also don’t chase perfection
  • 42:14 – You can find Amit on Twitter at @ml_amit and on LinkedIn
📄 References
⚙️ Tech Stack
🛠 Libraries Used

Support the Show

This episode does not have a sponsor and this podcast is a labor of love. If you want to support the show, the best way to do it is to purchase one of my courses or suggest one to a friend.

  • Dive into Docker is a video course that takes you from not knowing what Docker is to being able to confidently use Docker and Docker Compose for your own apps. Long gone are the days of "but it works on my machine!". A bunch of follow along labs are included.
  • Build a SAAS App with Flask is a video course where we build a real world SAAS app that accepts payments, has a custom admin, includes high test coverage and goes over how to implement and apply 50+ common web app features. There's over 20+ hours of video.

Questions

May 11, 2020

✏️ Edit on GitHub