I am a tax lawyer.
You probably need years and multiple millions of dollars to write competent tax return preparation software. Our office uses an expensive pro package (Lacerte, owned by Intuit) which is frankly a C- product.
I don't know what would be more important -- coding skill or a deep, intimate knowledge of tax law. Tax law is an internally inconsistent logic box. Oh. And ambiguous too. I'm not sure how you deal with those situations.
Even if you succeed in writing awesome tax prep software, you have a marketing problem.
We do not like Lacerte. We have identified software we would rather use. But I cannot get my team to move to the new software. Learning curve fear, data conversion fear. Etc.
Look at the accounting software business. Why does Quickbooks continue to dominate the market?
If you are interested in breaking into the market, find a tiny subset that you can competently serve. E.g., my friend Mary Beth and her business partner solved a truly horrific piece of tax hell at www.form8621.com. You have no idea about the intellectual complexity of the tax law underpinning her software.