Thanks for taking the time to read the article.

Here are the answers to your questions.

1. My management team understood the fact that although I wasn’t a seasoned veteran in the AI industry, I was a good culture fit and had some experience with the specific computer vision they wanted to implement. In terms of how supportive they were, the support I received was the best I could have asked for, I was provided with the one thing I required, and that was time. They understood that ML is a difficult field and learning never stopped, so they put their trust in me in setting my own deadlines for specific tasks and projects.
2. I haven’t made any major mistakes, and hopefully, I make none. But I do make minor mistakes on a day to day basis, but the key fact is that I learn from each and every one of my mistakes. An example of a mistake I’ve made recently is when creating the technical specification for the company’s GPU workstation; I recommended a well-known GPU workstation and provider but I failed to check up on their delivery service and costs, and they were actually very expensive. I rectified my mistake by realizing the need to conducting my own research regardless of what the general community was and I supplemented the management team with 2 new cheaper UK based GPU providers.
3. Well, paperswithcode has a nice feature that ranks techniques and solutions to a particular machine learning technique based on accuracy. From the results provided I select one of the approaches that are within an accepted accuracy criterion. I have a quick read of the research paper and thereafter look into GitHub for any implementation of the technique/algorithm presented in the paper. TensorFlow and PyTorch already a number of state of the art models built into the respective library. I utilize TensorFlow Hub a lot lately to gain access to some modules for tasks such as image classification and object detection.

I hope the answers I have provided are useful to you and, feel free to ask me any further questions.