Learn how to debug production code and then explore real time machine learning

On Oct 12, join other pythonistas to learn how to debug your production code and then explore real time machine learning with celery and django. Yelp, our sponsor will be providing food and drinks!

Lightning talks:

"Estimators - a simple way to Version your Machine Learning Models" by Simon Frid

"Python unikernels" by Ian Eyberg

If you'd like to give a lightning talk, sign-up here. (https://docs.google.com/forms/d/1LtV839ktupRboMUSXlXqoJ9lLFvpe-TZtLCf2q6jUpY/viewform?usp=send_form#start=invite)

Talk #1: Everything is broken, and I don't know why

SUMMARY

We all have someone on our team who writes code with bugs in it. Not us, of course, but someone else. This code somehow gets into production and this code just maybe sometimes crashes. How can we detect these issues and extract useful information at runtime so we can fix these bugs?

In this talk, attendees will learn about how to introspect their python application during a crash and learn how to extract more useful information than just a traceback to aid in debugging production issues.

BIO

Matt Robenolt is a software engineer at Sentry, previously operations team lead at Disqus.

Talk #2: Synchronous Celery, Django, and Real-Time Machine Learning

SUMMARY

Using Celery with Django probably makes you think of kicking off asynchronous tasks to run in the background. At Boomerang we've also found Celery to be a great option for managing heavyweight worker processes that process data in real-time. Building machine learning and image generation microservices on this architecture allowed us to build performant, scalable solutions for millions of users.

DESCRIPTION

You've picked your platform, built your models, loaded your training data, and optimized your algorithms. Now it's time to integrate your solution with your Django app. But how do you deploy machine learning in a way that you can run the algorithms fast enough to satisfy your users, scale appropriately and protect your application servers from crashes?

It turns out that using Celery synchronously provides a performant, robust, isolated, well-understood, and nearly-infinitely scalable method for deploying machine learning (or any other kind of technology that needs heavyweight workers) microservices.

In this talk, I'll show you how we've used this architecture twice at Boomerang, and how it evolved from generating calendar availability images in real-time to deploying worker processes that have already loaded deep learning models and can run algorithms against new data as soon as it arrives. This talk will go through our integration of Celery with machine learning systems and Django. I'll share some caveats and surprises that we encountered along the way and provide some general advice for setting up your servers and deploying into the cloud. I'll discuss scaling to millions of users as well as reliability issues we encountered and how we overcame them.

BIO

Alex is the CEO of Baydin (Boomerang), an email productivity company that combines artificial intelligence and behavioral science to ease the burden on overloaded emailers. Alex only gets to write code a couple days a week, but those are his favorite days. He graduated with Master’s and Bachelor’s degrees in EECS from MIT. When taking a break from his email, Alex likes to cook southern and southeast Asian food and is a rabid Alabama football fan.

Agenda:

6:00p - Check-in and mingle, with pizza and beer provided by our generous sponsor Yelp!

7:05p - Welcome

7:10p - Talk #1 and Q&A

7:50p - Announcements and lightning talks

8:10p - Talk #2 and Q&A

8:50p - More mingling

9:30p - Doors close

Please note these important check-in details at Yelp:

1. Doors open at 6:00pm to allow enough time for the check-in process. Before 6:15pm, please wait outside without blocking the building entrance. Attendees on the waitlist will be admitted at 6:45pm, after all attendees who RSVP'd have been admitted. Doors close at 7:30pm.

2. Please update the name on your Meetup account to reflect your true FIRST NAME and LAST NAME. Yelp security will be checking IDs downstairs. If your name on Meetup.com is not the name on your ID, then please enter your full name here (https://docs.google.com/forms/d/1d_oPoxjcAQzOJqozHIzVuFNnOYi7CDrzouywq4U9SUo/edit).

3. Since alcohol will be served at the event, we ask that any underage attendees RSVP directly to the meetup organizers.

4. Unfortunately, Yelp cannot safe-keep your bicycles. Please park your bike on the street or in the bike rack in the Fifth & Mission Parking Garage (https://www.google.com/maps/place/The+Fifth+%26+Mission+Parking+Garage+%2F+Yerba+Buena+Parking+Garage/).

**SF Python is run by volunteers aiming to foster the Python Community in the bay area. Please consider making a donation (https://secure.meetup.com/sfpython/contribute/) to SF Python and saying a big thank you to Yelp for providing food, drinks, and the venue for this Wed's meetup.