GPU Utilisation in Kubernetes, Open Telemetry, & Coding Coventions

Please join us for a fantastic PyLadies London summer meetup in collaboration with JPMorgan Chase & Co.

For security reasons, we are not openly disclosing the venue address. The nearest tube station is Blackfriars. The exact address will be sent to attendees who have signed up for the event prior to the event taking place. You must sign up with your legal first + last name or you will be denied entry to the building.

📝 Agenda:
• 6:30pm Doors open, socialising, food and drinks offered by our sponsor
• 7:00pm Talk 1: Maximizing GPU Utilization in Kubernetes: Strategies for High-Performance Python Applications
• 7:15pm Talk 2: Telemetry with monkey patching & auto-instrumentation
• 7:30pm Talk 3: The coding conventions that make our lives easier
• 8:00pm More socialising
• 8:30pm Fin

Talk 1, Mia Walls (Vice President): Maximizing GPU Utilization in Kubernetes: Strategies for High-Performance Python Applications
Abstract:
In certain high-performance computing (HPC) scenarios, GPUs can remain underutilized, posing concerns due to their associated costs. To address this, it becomes crucial to explore methods that maximize GPU utilization. In this talk, I will discuss utilization improvement strategies specifically within Kubernetes deployments. I will also present a practical example of how a Python application could be structured to maximize GPU utilization, particularly for workloads that do not fully saturate the GPU.

Talk 2, Mary Louise Reyes (Senior Associate): Telemetry with monkey patching & auto-instrumentation
Abstract: OpenTelemetry python library can be used by large applications to generate and collect telemetry datapoints about internal operations.
We use this for auto-instrumenting Python applications which enables us to collect traces end-to-end execution of a single transaction, as well as collect metrics - aggregated data relating to multiple transactions or a service as a whole.
For us in JPMorgan, our goal is to have unified telemetry data. We'll talk about what is and why we use instrumentation for our large software applications. We'll discuss Monkey patching and how we wrap un-instrumentable functions in custom callbacks defined in OpenTelemetry instrumentation library.

Talk 3, Çağıl Uluşahin Sönmez: The coding conventions that make our lives easier
Abstract: Engineers review code as often as they write it and use a lot of tools to improve this process, such as linters and auto-formatters. One tool that is also very useful is the coding conventions. It is a collection of references that can be used during code-reviews, whilst also serving as a valuable day to day reference and guide for all team.
This talk will explore several different conventions from open source conventions repo, offering an in-depth exploration of how they improve the quality of the codebase and serve as a tool for knowledge transfer among engineering teams.

The structure of the talk will follow:

* What are coding conventions and why do we use them?
* How to establish your own coding conventions?
* Examples of some of the most useful coding conventions and how they enhance code quality.
* How to uphold your coding conventions with the help of linters?

Attendees will leave with a comprehensive understanding of the concept coding conventions and how to implement them within their own team.

Speaker Bio: Çağıl is a software engineer based in London. She has been working with Python and Django since 2008. Currently she is working as a Lead Backend Engineer at Kraken Tech. She is vice president in Django Software Foundation, co-organiser of London Django Meetup, Pycon Turkey, DjangoGirls Turkey, and a PSF managing member.

🚨 Our meetups are governed by a Code of Conduct. Please take a few minutes to read it.

🥰 This meetup is sponsored by: JPMorgan Chase & Co.