Tokens & Tuning : LLM Happy Hour
Tokens & Tuning | Hosted by Linksia & BlueSkyCompute
A cocktail night for builders behind the models
✨Fine cocktails, sharp minds, and even sharper models.
🧠Curious about how other builders are running LLaMA and Hugging Face workloads off-cloud with faster throughput and lower cost?
👋 Let’s talk at the bar.
Expect sharp conversations, exceptional drinks, and peers who care about throughput, customization, and escaping GPU pain.
No slides. No stage. Just real builders, real conversations, and the occasional tensor joke. 🤖
🍸What to Expect
Quiet corners for talking fine-tuning, vLLM, LLaMA-3, or compute scaling
People who know what a flash-attention patch actually means
Good people who build serious models — and know GPUs aren’t cheap
Color-coded areas by topic — each with someone from the field to keep the conversations flowing
⚡Who Should Come
Engineers deploying or adapting large models
Founders building AI-native products
Researchers in language, vision, or multimodal domains
Builders navigating scaling issues or GPU cost ceilings
Infrastructure experts looking to optimize cost/performance
Anyone who’s debated 4-bit vs 8-bit quantization over drinks
Need more control over your model stack? Curious how others are escaping spot interruptions and GPU markups?
There’s no pitch here. But if you’re tired of getting throttled on AWS, someone in the room may have answers.