Tensor flow Reinforced Core for Metatrader 4/Financial Trading Exploration

I am looking to work with an experienced Tensor flow programmer to explore financial trading strategies within MT4. I will create multi input data environments within MT4 with MQL4 code – you will build an AI module that reads the data and provides an appropriate output response for placing market orders. You must be able to advise on data configurations around inputs I wish to test in the hope that a winning strategy manifests through machine learning. We will try reinforced learning first and maybe also try supervised learning. You must also provide full source code and be free at weekends/odd evenings for conference calls which will also cover explanations of programming AI interface so that I may become familiar with Tensor flow application coding myself.

Spec:
• Language C++ (preferably visual studios).
• All data inputs will be passed via pointer addresses using pipes (through DLL).
• All outputs will be command strings and will again be sent via pipes.
• Method of work: £ rate p/h, estimated to strict p/item cost – including part data reconfigures where applicable to keep overall project costs in control (negotiable/pph only).
• Multithreaded workload considerations for parallel trials via separate MT4 installations.
• Short scale testing may be done by AI programmer; long scale tests and learning runs will be done by me on one or both of my MT4 rigs.
* Saving and loading of data sets (agents).
• Memory usage should be around 4GB for each AI core installation (keeping in mind MT4 requires 1GB, so I can run between 1 and 4 learning cycles at a time).
• I will do both back tests and forward tests to assess if a strategy should go live or not.
Providing we can agree a rate that makes the project look viable, I will consider giving a greenlight, up until that point we can explore its merits via email and Skype etc.