California and Bengaluru-based Jovian, a platform for data scientists to track and reproduce machine learning experiments, collaborate with teammates, and automate repetitive tasks, announced it has raised $450,000 led by Arka Venture Labs. The other investors who participated in the round include Better Capital, SenseAI, Axilor Ventures, and other individual angel investors from Silicon Valley.
Commenting on the investment, Radhesh Kanumury, CEO and Managing Partner of Arka Ventures said,
“With the proliferation of data scientists across the globe and Jovian providing DevOps capability in that area, it is a great space to be in.”
Owned and operated by SwiftAce Inc., Jovian was part of the Axilor Accelerator programme in 2018.
“Based on our experience and also of our portfolio firms, we had seen a huge gap for continuous collaboration between dispersed data science teams as well access to models under experimentation. We feel Swiftace is well positioned to be a true interactive platform for data scientists,” said Vinish Kathuria, Managing Partner, SenseAI.
The startup, founded by Aakash NS and Siddhant Ujjain in 2018, plans to use the funding to grow its engineering team and further develop the product. It is also looking for community development via meetups, webinars, online courses and hackathons, and customer development in India and the Silicon Valley.
Jovian is building the tools, workflows and collaboration stack to power the future of artificial intelligence (AI) and machine learning (ML). As per the startup, the platform is language, framework, and cloud-provider agnostic, and easy to try out, and it tracks everything (datasets, source code, hyper-parameters, trained models, etc.) in a simple yet powerful online dashboard.
Blume VC, Benhamou Global Ventures and Emergent Ventures launch B2B cross-border accelerator pl...
Speaking about the venture, Aakash NS, CEO, Jovian, said,
“Data science is fundamentally different from software development, as it is much more experimental in nature, and requires you to keep track of many things like datasets, Jupyter notebooks, models, parameters, metrics, etc. apart from the source code. This is currently a very cumbersome process, which hinders productivity and collaboration. We’re trying to change that for the better.”
Using the platform, teams can also discuss and collaborate on their work, and create pipelines to automate analysis and evaluation of models.
“We’ve worked as data scientists for a couple of years, and as software engineers for several years before that. That’s how we realised there’s a huge gap in the quality of tools available to data science teams for managing their work. We started out by trying to solve our own problems, and soon found a lot of interest from the community in tools we were building," added Siddhant Ujjain, CTO, Jovian.
“We have also received interest from data science teams looking to use Jovian for collaboration within their companies, with a willingness to pay for more advanced features. We are currently in the process of on-boarding our first few paying customers,” Aakash told YourStory.
Going forward, Jovian aims to become the de-facto tool for the data science community to share and collaborate on data science projects online, similar to what platforms like Github have done for open source software.
“Our aim is to make Jovian really easy to try out and use for first time users, while at the same time we plan to provide all the advanced features large enterprises might require to manage their entire company’s data science work,” said Aakash.
“While there are many companies that offer cloud-based graphics processing unit (GPU) infrastructure for running machine learning jobs (e.g. AWS Sagemaker, Google Colab, Azure ML Platform, etc.), we do not compete with them, rather integrate with them and complement them by offering an experiment tracking and collaboration layer on top of these offerings,” he added.
Globally, Jovian competes with the likes of Comet.ml, Weights & Biases, and NeptuneML, among others.
“What sets Jovian apart is that it’s really easy to get started, it works in any language, framework, and cloud-provider, and also has a great support for Jupyter notebooks, a popular IDE (Integrated Development Environment) for data science,” Aakash said.
The startup is targeting a global audience of data scientists, machine learning engineers and AI researchers. While this is an emerging market today, it is expected to balloon to millions of users over the next few years, since many startups and large enterprises are increasing their focus on AI, and setting up in-house data science teams.
“Data Science will be at the heart of everything in the future, and we are likely to have more data scientists than software engineers. In light of that, building productivity and collaboration tools for the new world of data science presents a massive opportunity,” said Vaibhav Domkundwar of Better Capital.
Mumbai and US-based Arka Venture Labs has also backed startups like Revvsales and Primaseller previously. Both the startups are based out of the US.
Arka primarily invests in B2B startups that have a minimum viable product (MVP) with a strong potential in the US or in Europe. The new accelerator platform-cum-fund was started with a corpus of about $6 million (Rs 40 crore) with all three funds - India’s Blume Ventures and US-based two VCs (BGV and Emergent Ventures) anchoring the investment.
[Funding alert] Data science startup TheMathCompany raises Series A round from Arihant Patni