> This u200d seems to be a zero width space.
It's a [zero-width joiner][1], actually. [A zero-width space is something else][2].
> convert a unicode entity to an utf-8 character?
All UTF-8 characters *are* Unicode entities.
Are you perhaps saying you have a literal "u200d" string (5 ASCII characters) in the input and need to turn it into an actual Unicode character, encoded as UTF-8? If so, then presuming the terminal locale is UTF-8:
----
``` shell
$ sqlite3 :memory: "select replace('अमिताभ बच्u200dचन', 'u200d', char(0x200D))"
अमिताभ बच्चन
$ !! | od -t c
0000000 अ ** ** म ** ** ि ** ** त ** ** ा ** ** भ
0000020 ** ** ब ** ** च ** ** ् ** ** 342 200 215 च
0000040 ** ** न ** ** \n
0000046
```
----
That "342 200 215" bit is the octal representation of the character 0x200D encoded in UTF-8.
In other words, you might be looking for a combination of [`replace()`][3] and [`char()`][4].
**EDIT:** Shortened example code for clarity. Also, [bang-bang][5]!
[1]: https://en.wikipedia.org/wiki/Zero-width_joiner
[2]: https://en.wikipedia.org/wiki/Zero-width_space
[3]: https://sqlite.org/lang_corefunc.html#replace
[4]: https://sqlite.org/lang_corefunc.html#char
[5]: https://unix.stackexchange.com/a/3748/138