Hi, I’m looking to hire a rasa dev so that I can build out my own platform and move away from third party hosted solutions. What skillsets should I look for? Thanks!
Two things If you are building bots for other businesses, at this stage in order for you to provide better service in terms of which pipeline to select and how to manage the intent classification and how to use core properly, you have two options - Either
Rasa Enterprise which comes with a platform and I suppose support from ML engineers/AI Researchers knowing the product ( This ain’t a sales promotion for them 
)
You will still need an ML expert to be able to help your clients understand their use cases, something I suppose you already do with third part solutions and also a devops expert to be able to put them into your infrastructure
OR
Hire an engineer (Full stack) to build out the platform using the frameworks (Angular/React/Vue) out there.
A Data Scientist/ML Expert( could be even part-time) to help your clients understand the intent classification and also how to design the conversation with them. It is quite important that you will need someone with an understanding of how exactly the rasa pipeline works and how it behaves when it comes to different classification techniques that are out there. Keep in mind Rasa is able to put all these different NLP techniques ( Classification, NER and Dialogue Management) together and stitched them out to a product. We have two roles (ML engineer/Data Scientist) in our organisation helping our internal businesses understand their own use-cases. The ML engineer usually helps the DS stitch the scripts together in order to ease the training and deployment of models
A Devops expert hosting the entire platform on infrastructure of your choice.
This is our team formation. You don’t need to find a Rasa expert, you need the person to become one. Solid understanding of computer science and along with good IT skills is what we look for 
@souvikg10What are you doing in your company with ML exactly? Do you just test different features for the algos or do you create new algos?
We are a retail bank. So it is mostly applicative. At the moment we try out different features suited to our problem case most of which is related to NLP.
however there are other types of advanced statistical modelling that has been done in banking for years specially in Risk Assesment, Mortage scoring, Fraud etc etc.
but more recently we are looking at ways to use Machine Learning at Scale directed towards our clients like Real time Fraud detection, Conversational AI, Email classification . Like less than a year recent. and on the subject of Conversational AI , organizational structure is quite often a subject of discussion for us. Building a chatbot is one thing but managing it at scale is a whole different subject.