MatX jobs

We build the world's best chips and racks for training and serving large language models. Our work spans the entire range from silicon and rack design, through compilers and kernels, to ML models.

We believe, more than any other AI chip company, that low precision compute performance is the most important aspect of chip design.

- MatX Values -

Consider new approaches

We're solving a different problem than previous products, so our solutions will often be different. We readily abandon traditional approaches when there's a better path in our context.

Understand deeply

We carefully reason through consequences of big decisions. A few days of thinking can be worth more than weeks of simulation. This is how we make big bets with high confidence.

Invite and include

We're a high-trust team who supports each other. We take contributions seriously and make sure good ideas are heard, even when they mean major shifts in direction.

Ship quality work

Whether in hardware, software, or ML, we value readable, resource-efficient code and solutions to real technical problems. We're interested in all experience levels.