Deep Learning passionate iOS developer to further build a simple Tensorflow image classifier and deploy it on our existing App

On the lookout for a bright image processing enthusiast to develop the image recognition side for an existing iOS-only mobile application - build a simple Tensorflow image classifier and bring it in production mode. You should create and train a neuronal network using the pre-trained Inception model (we already have a head start on this, have great results in the model selection and training, but might add a couple of classes and images) and do transfer learning starting with 500 classifiers for which we have around 1'000 training pictures each. Then we have testing data (pictures taken with the phone for the 500 classifiers) that can be used for retraining. For this phase, we might only need some fine tuning for the model we already have, improve deep learning algorithms to improve our current classification performance. Tensorflow returns the best results from our initial tests. Luckily, with TensorFlow, deep machine learning transitions from an area of research to mainstream software engineering that requires minimum coding. Until now, a couple of lines of Tensorflow Python, and a bag of "tricks of the trade" have done the job. Then, simplified, our app would send the picture(s) in the Amazon AWS Cuda instance where this deep-learning network prediction API does the hard part in recognizing the object​ and then sends back the text results to our app's API, shows matching classifiers to the user and the user just selects just a tag out of the first 5 results (if that's the case, correct recognition). There's also the option of implementing Tensorflow mobile, on-device processing for iOS [login to view URL] For the future, we should also consider enabling continuous training from the users' input. Prefer somebody with a strong passion for the latest developments in deep learning, rather than an "industry expert", therefore experience in similar projects is not mandatory. Like this video shows, it's not a matter of having a PhD anymore: [login to view URL] You should be passionate and eager to play with TensorFlow and have the ability to bring this to production. We already have the image classification in production, but currently using the Google Vision Cloud API, which doesn't render great results. You’ll not only be able to mention this experience in your C.V., onto Linkedin or your freelancer portfolio but also have the opportunity to work with a trending startup, where your input and creativity would be valued and encouraged, on a topic that is actual and that you're passionate about. QUALIFICATIONS Techie with a sparked interest in deep learning, data science, a striking passion for developing and expanding deep learning technologies and a basic foundation to get up to speed with: - Python; - Familiarity with DL libraries such as TensorFlow. - Technical proficiency GPU programming using CUDA. - Ability to deploy/ run ML codes in the cloud – AWS CUDA instance, we have Amazon credits. - Agileness and willingness to work independently and be self-driven.