PyData Cornwall - Meetup #5

🍂 As we tuck our collars up and move through Autumn, let's gear up for the fifth PyData Cornwall meetup on Tuesday 24 November! A lot of us have been busy building over the summer so we'd like to hear about what you've learned and explored over the past few months.

🗣️ Please let us know if you'd like to give a 30 minute talk or a 5 minute lightning talk (we might mix this up if we get more lightning talk submissions). First time speakers are very welcome and we're happy to help you prepare or practice a talk. Submit a talk here. If you have questions, best thing to do is jump on our Discord and say hi!

Venue
Flowmoco, 11 Trevena Terrace, Newquay TR7 1LJ

Agenda
🍕 Doors open at 6.30pm with drinks, talks start at 7:00pm, pizza arrives at some point and when talks finish we can move to a nearby pub for drinks.

🎙️ Main Talks (25 min + 5 min Q&A)
1️⃣ Beyond the Model: Jointly Optimising Hardware, Engines and Harnesses for Local Inference — Savio Rozario.
Running LLMs on your own hardware gives you privacy, control and predictable costs. But default setups are often slow and frustrating. The reason is usually not one weak part. It is how the parts work together. This talk looks at the whole stack: hardware, inference engine, model, gateway, harness and observability. It shows how tuning these layers together can make a modest machine fast and useful. I start with the key ideas you need, such as why memory speed limits output and what you lose when you shrink a model. I then share results from tuning a real setup, including what helped most and what barely mattered, along with settings you can reuse. I finish with the economics: when running models locally pays off compared with using an API, and which tasks it suits best.
2️⃣ The Grammar of Graphics - Bringing Sanity to Data Visualisation — Tariq Rashid.
The predominant tools for data visualisation in the python data science ecosystem is matplotlib. It is a very flexible tool, capable of a very wide range of charts. However, even experienced data scientists find it painful to use. This is primarily because the library enforces very little conceptual structure, allowing us to hack together code "until it works". In contrast, the Grammar of Graphics is a conceptual framework for data visualisation. It is implemented most successfully in R by ggplot, but python implementations are gaining interest. The key benefit of a firm conceptual framework is that code has a predictable structure, is easy to write and also easy to read. This beginner-friendly talk introduces the key concepts of the Grammar of Graphics, and illustrates them briefly with examples.

⚡ Lightning Talks (5 mins)
1️⃣ Learnings from Conferences (Part 1) — Ben Whorwood & Kevin Bath.
2️⃣ Learnings from Conferences (Part 2) — Fei Phoon & Tun Shwe.

🤙 We look forward to seeing you for another fantastic evening!

Notes
☑️ Please update your RSVP if you can no longer make it so we can estimate food and drinks. We have limited space, so be a pard and free up your place for other fellow community members. If you have any allergies or dietary requirements, please message us and we'll see what we can do.
👩‍🏫 We're a humble chapter of the global PyData community, and as such we are governed by the PyData Code of Conduct. To discuss any issues or concerns relating to the code of conduct or the behaviour of anyone at a PyData meetup, please get in touch with the organisers, or contact NumFOCUS Executive Director Leah Silen (+1 512-222-5449; [leah@numfocus.org](mailto:leah@numfocus.org)).
💸 We welcome sponsors so please get in touch if you can introduce someone that would like to sponsor pizza or drinks for the community or host a future meetup (for a dedicated pitch and speaker slot).
🪩 Join our Discord server to make friends, meet speakers, get help with your talks and be the first to hear about opportunities. We are not-for-profit run by tech professional volunteers, so our space is ad- and marketing-free, so come help make the free learning and networking community you want to enjoy!
🏴‍☠️ Follow PyData Cornwall on LinkedIn for updates and to share your thoughts and photos.