Ango Hub

Hi everyone! Gökalp here, co-founder and CTO of Ango AI 👋 After working with ML in production for 4 years, I realized the biggest obstacle to shipping quality ML-based products in time and within budget was data labeling. There are great open source and commercial data labeling platforms out there, but each has its own set of drawbacks. Me and my co-founder and brother @gokhan_urul wanted to make the missing data labeling platform. So we founded Ango AI (https://ango.ai) and set out to make Ango Hub, now used to annotate millions of data points by and for dozens of customers. Ango Hub fills what we felt were the biggest gaps in the industry. Here are some of its key features: 🌟 Versatility: It’s versatile, supporting any data type you throw at it. It handles image, video, audio, text, PDF, and multi-page images, all with a common interface so annotators don’t need to retrain. Plus it has nearly twenty labeling tools, from the common bounding boxes and polygons all the way to label relations, rotated bounding boxes, tables, and more. 🌟 Scalability and collaboration: Scaling a labeling project means managing lots of people with different roles. This is why Ango Hub comes with features such as intra-labeler consensus, review workflows, centralized instructions, a real-time issue/feedback system, a sample label library for your annotators, and more. On Ango Hub, labelers, reviewers & project managers aren’t just isolated silos, they actively collaborate and engage with each other. 🌟 AI Assistance: Nowadays you cannot imagine a labeling company without AI assistance. However, our experiences and studies showed that too much support from AI Assistance may result in a tendency to over-rely, besides most of the AI Assistance tools are domain specific and costly to use. In Ango Hub we focused on cost effective, domain agnostic, easy to use interactive tools to support labelers with the right amount. 🌟 Plugins: Even though most of the labeling platforms can cover the basic needs of labeling projects, small customizations on the platform can drastically improve the whole labeling process. This is why we created plugins. With plugins you can connect your own models, formatters and interactors easily. 🌟 Security: It was built with security in mind from day one. While you can access Ango Hub from https://hub.ango.ai, import your assets securely, and start labeling immediately, we have plenty of options for when a complete lockdown of data is needed, including the possibility of installing Ango Hub directly on your own datacenter, on-premise, so that your data never leaves your server. Installation only lasts 15 minutes on average and is Kubernetes-based. 🔥We also found that existing on-premise data annotation platforms were only affordable to those who could dedicate massive budgets to labeling. With Ango Hub, we want to democratize and open up on-premise data annotation to midsize teams. It’s now installed and used in companies ranging from medium-sized startups all the way to national banks. Plus, everyone who signs up today gets a free plan that should be enough for small teams to get even medium-sized projects done. I can’t wait to hear your feedback and thoughts 🕺 On a final note, I'm super proud of the Ango team who worked really hard on this product and launch. Kudos 🙏

Congratulations @gokalpurul, @gokhan_urul and team. I remember working on a project which required labeling data in images, videos etc. This will not only solve sub-problems of such type but you have provided so many features on top of it that it will complete the end to end flows of the problem. I can see a good scope for Ango Hub, all the best.

Amazing, you are solving one of the biggest problems for folks who want to train ML models! Thanks!