Lamb Labs

World's fastest AI models

Y Combinator

http://lamb-labs.com ↗

Jobs Hiring Discussion

Company facts

Early-stage

Founded: 2026
Headquarters: United Kingdom

About

Lamb Labs is building custom chips for AI inference and the world's fastest LLM model at the lowest power. We've developed a new post-training technique that converts existing models to a diffusion-based architecture that delivers 2× faster inference on existing GPUs. We then build the hardware, custom silicon chips targeting up to 20,000+ tok/s and a 63× higher intelligence per watt.

1 job

Posted Sort

1 job Posted Sort

Lamb Labs

yesterday

Member of Technical Staff (Model Quantization & Inference)

NewBe an early applicant On-siteUK

United Kingdom

ML engineer/researcher at Lamb Labs developing extreme low-bit (1–2 bit) quantization and inference methods for large language models — covering weight/activation quantization, KV-cache and attention optimization, and QAT/PTQ/distillation — to make models run efficiently on the company's custom AI chip (MPU), working directly with FPGA/ASIC engineers. Core tools: Python and PyTorch.

AI CUDA Fine Tuning FPGA GPU +7 skills

Apply