These Chinese op-amps have 1fA input bias currents and only cost 0.1715$
https://lcsc.com/product-detail/Low-Power-OpAmps_3PEAK-TP2121-TR_C248708.html
so the questions is How it's even possible when the switch leakage currents are around 1nA and the op-amp input bias currents are around 20fA, how they claim 10fA range!
How the design can be improved?
There's nothing to improve. It's a potentiostat controller for cyclic voltammetry and, as the ample literature indicates, is fit for purpose. It does exactly what it needs to do for the scientist to do the experiment. Read the articles I found for you and you'll quickly realize it's more than just something to measure current.
If you want to use this for something it isn't design for, you'll probably be disappointed.
You have now entered the realm of science and left the world of engineering. The electronics are merely a tool to get the data. As long as you know the limitations of the tool, you accept it as suitable. If you don't understand the scientific application, you will find faults with the design and claim it needs improving. It doesn't. It does the job.
The authors make it very clear that high input capacitance is an issue and they demonstrate that oscillation occurs - BUT IT DOES NOT MATTER for their application for the reasons they explain. It would matter if you were trying to use it for a more general purpose application.
When I was a PhD student developing a lab instrument, if I had listened to the electronics engineers employed by the university, I would never have built it. They said things I was trying to do were impossible and they refused to help. Well, my impossible idea is now installed in thousands of labs around the world and a number of instrumentation companies sell their versions of it. In spite of their millions of dollars and professional engineers, my 34-year design still runs circles around theirs. But it is fine-tuned to do one thing and do it well, not be a general purpose tool.