What you have to do is not an encoding conversion but a custom character mapping.
In an encoding a specific byte or byte sequence stands for a specific character. The font then visualizes this character. For example, in ASCII the byte x41 stands for the character "A", and different fonts have different shapes to display this "A" visibly on screen.
In the case of Kruti Dev, apparently at the time it came into being, there was no encoding for Indian languages; i.e. there was no particular byte specified that should represent "व" in any system in use at the time. What the creators of Kruti Dev did was simply redefine the shape of a letter. The bytes still said the letter was "A", the operating system still thought it was handling the letter "A", but the font contained the shape of "व" for visual display.
So there's no encoding conversion you can do here, since the underlying encoding is being abused in non-standard ways to begin with. What you need to do is to map latin letters to actually specified Indian letters. You need to substitute every "A" for "व" (just an example, no idea about the actual mapping).
