talk-data.com

Details

Venue: Riverbank House, 2 Swan Ln, London EC4R 3AD

Please note:

1. 🚨🚨🚨A valid photo ID is required by building security. 🚨🚨🚨
2. This event follows the NumFOCUS Code of Conduct, please familiarise yourself with it before the event.

If your RSVP status says "You're going" you will be able to get in. No need to show your RSVP confirmation when signing in. If you can no longer make it, please unRSVP as soon as you know. *** Code of Conduct: This event follows the NumFOCUS Code of Conduct. Please get in touch with the organisers with any questions or concerns regarding the Code of Conduct. *** As always, there'll be free food & drinks, generously provided by our host, Man Group. ***

Main Talks

1. Death by RMSE: A Cautionary Tale of Metrics Gone Wild - John Sandall You’ve trained the model. The eval metrics look great. But somehow, it doesn't change anything — the KPIs are static, the business impact isn’t there, and you’re left wondering: did we optimise for the wrong thing? In this talk we explore how the wrong model evaluation metric can quietly sabotage your machine learning team's impact, and hear my own first-hand experiences of where the choice between MAE, F1, RMSE and others wasn't academic but business critical. We'll cover the power of cohorting, backtesting to simulate past futures, how to align eval metrics to downstream company KPIs, and some stories of teams that got it all disastrously wrong. Whether you’re building models, leading teams, or just quietly suspicious that your ROC-AUC obsession might be misleading you… this talk is for you.

John is CEO of data science consultancy Coefficient, and CTO of stealth AI marketing startup Day30.

2. LLMs for Recommender Systems - Janu Verma Large Language Models (LLMs) emerge as promising tools due to their world knowledge and reasoning capabilities, enabling nuanced understanding of user preferences and contextual patterns. In this talk, we will discuss how these models—designed primarily for language tasks— fare in recommendation tasks. I'll walk through my research on LLMs for recommendation problems. We will investigate pre-trained LLMs as zero and few-shot recommender systems as well as fine-tuning LLMs for recommendation tasks. We will discuss how to build a movie recommendation system by supervised fine-tuning a small-scale LLM on user interaction data.

⚡ Lightning Talks 1. What even is "\" in Python?. - Alex Caian

2. Beyond Conversation: Building AI Assistants That Take Action with Model Context Protocol - Amaboh (Ama) Achu

Logistics Doors open at 6.30 pm (get there early as you have to sign-in via building security), talks start at 7 pm, drinks from 9 pm in the bar. We will have reduced capacity for this event but there will be plenty of people to discuss data science questions with! Please unRSVP in good time if you realise you can't make it. We're limited by building security on the number of attendees, so please free up your place for your fellow community members! Follow @pydatalondon (https://twitter.com/pydatalondon) for updates and early announcements.

Join us at Newspeak on September 24th for an evening of talks, food, and conversation with ML and AI industry pros.

Please note you will be unable to enter the venue before 6.30pm.

RSVPs will close 24 hours before the event, you may be unable to register after this time but you can still watch online.

If you can't join us in person you can watch remotely via our YouTube channel.

Agenda

06:30pm - Doors open, food and drink served

07:00pm - Welcome

07:05pm - A short talk from our sponsor Daemon

07:10pm - John Sandall, CEO & Principal Data Scientist at Coefficient "Fairness Tales: Investigating the use of AI tools in hiring and recruitment"

07:50pm - Break

08:00 - Andreas Varotsis, Head of AI Capability at 10 Downing Street "Measuring Bias and Impact in GovAI"

09:00pm - Wrap up

Please RSVP for the event well in advance if you plan to attend in person and unRSVP if you can no longer attend as limited spaces are available.

Venue: Carnival House, 100 Harbour Parade, Southampton, SO15 1ST 📢 Want to speak 📢: submit your talk proposal

Please note:

1. 🚨🚨🚨A valid photo ID is required by building security. You MUST use your initial/first name and surname on your meetup profile, otherwise, you will NOT make it on the guest list! 🚨🚨🚨
2. This event follows the NumFOCUS Code of Conduct, please familiarise yourself with it before the event.

If your RSVP status says "You're going" you will be able to get in. No further confirmation required. You will NOT need to show your RSVP confirmation when signing in. If you can no longer make it, please unRSVP as soon as you know so we can assign your place to someone on the waiting list.

*** Code of Conduct: This event follows the NumFOCUS Code of Conduct, please familiarise yourself with it before the event. Please get in touch with the organisers with any questions or concerns regarding the Code of Conduct. *** There will be pizza & drinks, generously provided by our host, Carnival UK. ***

Main Talks 1️⃣ The Python Powered Modern Data Stack - John Sandall The Modern Data Stack has brought a lot of new buzzwords into the data engineering lexicon: "data mesh", "data observability", "reverse ETL", "data lineage", "analytics engineering". The PyData Stack introduced pandas pipelines powering ETL workflows...clean code, tested code, data validation, perfect for in-memory workflows. As demand for self-serve analytics grows, new data sources bring more APIs to model, more code to maintain, DAG workflow orchestration tools, new nuances to capture ("the tax team defines revenue differently"), more dashboards, more not-quite-bugs ("but my number says this...").

Learn why the world shifted from ETL to ELT, what open-source Modern Data Stack tools you should know, with a focus on how dbt's Python integration is changing how data pipelines are built, run, tested & maintained. Beyond the latest trends & buzzwords, there is value to be gained by understanding Python's role in the future of data engineering.

2️⃣ Transformers Inside Out (Part 1.5) - Sam Joseph Large Language Models like GPT4 have catapulted transformers into the limelight, but how do they really work? How can you really code them? This talk presents an alternative representation of GPT internals and a new set of flow diagrams that more closely match the code in order to better understand transformers inside out!

Lightning Talks ⚡ ⚡1️⃣ Estimating cognitive workload using electroencephalography - Christoph Tremmel ⚡2️⃣ Benchmarking Time-Series Databases - Nick Thorne

Logistics Doors open at 6.30 pm, talks start at 7 pm. For those who wish to continue networking and chatting we will move to a nearby pub/bar for drinks from 9 pm.

Please unRSVP in good time if you realise you can't make it. We're limited by building security on the number of attendees, so please free up your place for your fellow community members!

Follow @pydatasoton (https://twitter.com/pydatasoton) for updates and early announcements. We are also on Instagram/Threads as @pydatasoton, and find us on LinkedIn.

Venue: 1 Angel Lane, EC4R 3AB

Please note:

1. 🚨🚨🚨A valid photo ID is required by building security. You MUST use your full real names on your meetup profile, otherwise, you will NOT make it on the guest list! 🚨🚨🚨
2. This event follows the NumFOCUS Code of Conduct, please familiarise yourself with it before the event.

Tickets are assigned through a lottery draw about 1 week before the event.

If your RSVP status says "You're going" you will be able to get in. No further confirmation required. You will NOT need to show your RSVP confirmation when signing in.

If you can no longer make it, please unRSVP as soon as you know so we can assign your place to someone on the waiting list.

Code of Conduct: This event follows the NumFOCUS Code of Conduct, please familiarise yourself with it before the event. Please get in touch with the organisers with any questions or concerns regarding the Code of Conduct.

As always, there'll be free food & drinks, generously provided by our host, Man Group.

Main Talks 1️⃣ Evaluating Distributional Forecasts - Leonidas Tsaprounis

In this talk we will focus on the metrics you can use to evaluate distributional forecasts. Distributional forecasts can be in the form of monte-carlo samples of future paths or analytical distributions and they aim to describe the full distribution for a forecasted time point. Using simple implementations with numpy and scipy we will go through the properties of different metrics like the Continuous Rank Probability Score (CRPS) and the Log Score and use them to pick the best forecasting algorithm using cross-validation.

2️⃣ How To Build Your Own Private ChatGPT Using Streamlit, LangChain & Vicuna - John Sandall

Your garden variety Large Language Model (LLM) is pretty great but comes with some downsides: no information after September 2021, and potentially insecure if you want to leverage internal documents and sensitive data, so let's build our own Vicuna-powered “ChatGPT”!

Lightning Talks ⚡ ⚡1️⃣ Open Source is Technically Illegal – Casper da Costa-Luis You are probably breaking the law.

⚡2️⃣ London Data Week – Jennifer Ding London Data Week 2023 (https://www.londondataweek.org/) runs from 3rd-9th July, a citywide festival about data to learn, create, discuss, and explore how to use data to shape our city for the better. Find out what's happening across London, from Turing Institute's data cabaret to citizen science.

Logistics Doors open at 6.30 pm (get there early as you have to sign-in via building security), talks start at 7 pm, drinks from 9 pm in the bar. We will have reduced capacity for this event but there will be plenty of people to discuss data science questions with!

Please unRSVP in good time if you realise you can't make it. We're limited by building security on the number of attendees, so please free up your place for your fellow community members!

Follow @pydatalondon (https://twitter.com/pydatalondon) for updates and early announcements.