Lightning AI ⚡️

We’ve identified a security incident involving unauthorized access to the Lightning Platform involving unauthorized access to credentials.

This is an active investigation and during this time the Platform will remain inaccessible to users as our teams focus on addressing the event.

Please refer to our security notice: go.lightning.ai/4yOMOCd

PyTorch Conference starts next week and we’re bringing two sessions.

🔹 Turning Tribal AI Knowledge into a Model
Liana Koleva and Teja Pulagam tackle a question every ML team runs into: how do you turn what your best engineers know in their heads into something a model can actually learn?
📍 Room 210A | 🗓️ Wed, Oct 21, 11:45am–12:10pm PT

🔹 What Agentic ML Engineering Looks Like on Lightning
FNU Harsh and Sarah Ahmed break down what agentic ML workflows actually look like in practice.
📍 Demo Theatre, Community Expo | 🗓️ Wed, Oct 21, 1:30–1:40pm PT

If you’re at PyTorch Conference, come find us.

Join us at GTC Berlin next week.

On Wednesday, Oct 21, Lightning AI is on the NVIDIA × betahaus partner panel at the AI Startup Compute Night, talking GPU access, infrastructure, and what it takes to get compute at every stage, from prototype to production.

Then, stick around for food, drinks, and networking with AI founders, builders, and technical leaders.

📍 Berlin | 🗓️ Oct 21, 6–9pm CEST

Request to join → luma.com/6bx2qu5m

Join Lightning AI for a fireside chat on scaling RL on PyTorch, then stick around for a happy hour with a retro arcade and free food and drinks.

Sarah Ahmed (ML Solutions Engineer, Lightning AI) moderates with Alireza Shamsoshoara (AI/ML Engineer, PyTorch team @ Meta) on Oct 20th during PyTorch Conference, discussing why game environments are some of the hardest places to train RL agents: real-time constraints, huge action spaces, and players who notice the moment an agent feels off.

6:00 PM : Fireside chat
6:40 PM : Q&A
7:00 PM : Happy hour: retro arcade, free food and drinks

Register here → luma.com/3thl4zbw

Game environments are some of the hardest places to train RL agents: real-time constraints, huge action spaces, and players who notice the moment an agent feels off.

Join Lightning AI on Oct 20 during PyTorch Conference for a fireside chat on Scaling RL on PyTorch: Monarch, TorchTitan, and What Games Demand. Sarah Ahmed (ML Solutions Engineer, Lightning AI) moderates with Alireza Shamsoshoara (AI/ML Engineer, PyTorch team @ Meta)

6:00 PM : Fireside chat
6:40 PM : Q&A
7:00 PM : Happy hour: retro arcade, free food and drinks

Capacity is limited and registration is required for the address.

Register here → luma.com/3thl4zbw