Show HN: Hazlo.ai – An easy-to-use autoML platform

Hey HN, I'm Ojas from Neuralastic, we're building Hazlo, an autoML (or AIaaS as we like to call it) platform that allows anyone to train, deploy and host ML & DL algorithms (https://hazlo.ai ). Helping different companies integrate AI into their systems from my dorm room has opened me up to the diverse array of challenges that businesses are trying to solve with algorithms. Yet for SME or individuals without the resources or expertise to pull it off, it’s becoming harder to join this next industrial revolution. A couple of months ago, my co-founders and I got a hold of Google’s AutoML, which we though could significantly help users without expertise develop and host their own models. But, it turns out that to upload a dataset you have to create a bucket in the same region as your AutoML platform, then you have to clean, extrapolate and featurize your data, then it takes another solid hour to upload your dataset, after which you can only choose the prediction columns Google allows you too. If someone could do all this, they could probably deploy their own Keras model. Therefore, we decided on develop a completely cloud-based solution which would enable anyone to develop and use custom models on the fly. All one would have to do is upload their .csv or .xlsx, choose their prediction columns, press train and their model would be ready to go. And this is what Hazlo hopes to achieve. Building Hazlo has been an awesome journey, but we still have miles to go and a ton to learn. Our server power is a little shaky, the payments gateway is still being added, and the UI is not the best — but we still want to put it out there and possibly help someone get access to solid algorithms without funds or expertise in AI. We’ll also be adding an image-recog system which would allow users to upload images, type their categorizes and start using the model and just have a lot more in store. Of course, I’d love to hear any feedback or thoughts on the platform. Thanks!