PyTorch Meetup #21

Join us on September 12, 2025 to learn, share, and network with some of the most experienced professionals who are passionate about AI/ML.

This time, we’re excited to have Databricks as a special partner.

You’ll gain first-hand insights into the Databricks Cloud Platform and hear from our experts and partners across the industry.

Speakers

- Kir Zharov, Co-Founder at SpinnerAI:
"LLM Inference Under the Hood: From Edge Devices to Production Beasts"

- Udaybhanu Gaddam and Sanjay Ashok, Solutions Architects at Databricks:
"Automating LLM lifecycle on Databricks using Agent Bricks"

- Ash Vardanian, Founder at Unum":
"Scaling Vector Search Up & Out with Spark"

- Alexandr Notchenko, AI Research Scientist at Topnotch Labs:
"Foundation models for 3D reconstruction"

Event Details

Date: Friday, September 12, 2025
Time: 17:30 – 21:00
Where: Databricks London HQ
Location: 11–14 Windmill St, London, W1T 2JG

👉 Please fill out the registration form:
https://forms.gle/xkDkatBrSUWUdxaS6

Important: Only 120 seats available.
Registration is required. For security reasons, only invited guests on the official list will be granted access. Walk-ins will not be allowed.

Agenda

17:30 – 18:00 | ​Gathering, 🍕 & 🍺
18:00 – 18:15 | Welcome & Intro

18:15 – 18:40 | Kir Zharov: "LLM Inference Under the Hood: From Edge Devices to Production Beasts"
We will dive deeper into inference from Apple Silicon to huge production clusters, and cover tricks how to make it faster.

18:45 – 19:10 | Udaybhanu Gaddam and Sanjay Ashok:
"Automating LLM lifecycle on Databricks using Agent Bricks"
This talk explores what components are included in the LLM lifecycle: fine-tuning, evaluations, LLMs as judges, and human-in-the-loop feedback. We’ll show how Agent Bricks on Databricks automate these workflows, making it easier to build, assess, and scale trustworthy AI systems

19:10 – 19:20 | Break before two Lightning talks

19:20 – 19:30 | Ash Vardanian: "Scaling Vector Search Up & Out with Spark"
Apache Lucene was never built for AI-scale vector search. In this talk, I’ll show how USearch bridges the JVM–native gap, pairing Spark’s horizontal scale with native vertical speed.

19:35 – 19:50 | Alexandr Notchenko: "Foundation models for 3D reconstruction"
We will trace 3D reconstruction from classical SfM/MVS to the deep-learning shift, transformer-based models like VGGT that tackle multiple 3D vision tasks at once. This talk is for anyone at the intersection of deep learning and 3D vision who wants to understand how these tools are redefining the state of the art and the future of spatial AI.

19:50 – 20:00 | Start of Networking

20:00 – 21:00 | Continue of Networking in a pub (wait for the updates)

If you’d like to give a talk or become a partner for one of the upcoming meetups, message https://www.linkedin.com/in/fedorshabashev/ or https://www.linkedin.com/in/kirzharov/

In the meantime, you can watch the previously recorded talks:
https://youtube.com/@londonpytorchmeetup

Please subscribe to our channel! ❤️