In order to speed up machine learning pipelines, Strong Compute has raised a 7.8 million seed round

It was revealed today that Sydney, Australia-based firm Strong Compute has received a seed round of 7.8 million. Sequoia India, Blackbird, Folklore and Skip Capital are among the 30 funds and angels involved in the round. Y Combinator and Starburst Ventures are also participating, as are entrepreneurs and engineers from businesses such as Cruise, Waymo, Open AI, SpaceX and Virgin Galactic.

Y Combinator’s Winter ’22 batch included the startup, which claims that its improvements may speed up training by 10x to 1000x, depending on the model, pipeline, and framework. Since Ben Sands, co-founder of AR business Meta, formed Strong Compute, the team has recently achieved major advances where it was able to run Nvidia’s reference version, which its client LayerJot utilised, 20 times quicker. “That was a huge victory,” Sands added. We were left with the impression that “nothing is beyond repair.” There were certain aspects he didn’t want to divulge, but he emphasised that he and his colleagues are now employing mathematicians and developing tools to help them better understand how their user’s code interacts with the CPUs and GPUs.

Now that it has received this investment, the organisation can now get a head start on automating tasks that were previously done by hand in order to improve the training process. “Our aim today is to have major development partners in self-driving, medical, and aerial, so that we can be looking at what is truly going to generalise extremely well,” he said, referring to his company’s focus on developing autonomous vehicles. Our R&D team now has the resources to look at some genuine core technology that may take a year to truly win but that can really assist with the automated analysis of the issue, rather than just delivering something in a two-week sprint.

Over the next three months, Sands intends to treble the number of full-time engineers employed by the firm. Additionally, large corporations, many of which spend 50 million or more on computing resources, are now showing interest in the company. (And Sands noted that the market is essentially bi-modal, with customers either spending less than 1 million or 10 to 100 million, with only a few players in the middle).

This difficulty is shared by every organisation developing ML models: training models and conducting trials to enhance them still take a long time. There is a lot of time spent in a “holding pattern” for the well-paid data scientists working on these issues. Nikhil Abraham, SteadyMD CFO, remarked, “Strong Compute is tackling the basketball court challenge.” Devs spent their days shooting hoops and waiting for machines because of long training periods.

Strong Compute’s emphasis remains on computer vision for the time being, despite demand from the finance sector and firms looking to improve their natural language processing models.

“We’ve only just scratched the surface of what machine learning and AI can do.” said Folklore partner Tanisha Banaszcyk. “We love working with founders who have long-range ambition and visions that will endure across generations. Having invested in autonomous driving, we know how important speed to market is – and see the impact Strong Compute can have on this market with its purpose-built platform, deep understanding of the 500B market and world-class team.”