Tough TIA

Looks good, but... your input clamps can go through an inductor to ground instead of straight to ground; the input capacitor is assurance that the inductor doesn't saturate. That''ll help stretch out the protection diode currents when lightning strikes. Neon light would be best clamp for that kind of zap... put it outboard of C1.

And, the C2 is redundant if you drop the Q1/Q2 level translation (why use capacitors AND level translation?).

Also, if it matters, a duplicate resistor to R3 can go from (+) input to ground, just to keep input impedances DC-balanced.

I want this to be intrinsically safe in the outgoing direction too; the capacitance I'm measuring might be gasoline or something. Might need four diodes for super redundant safety.

C2 keeps DC out of the opamp, the actual TIA. I like the transistor because it should be very stable and isolates the opamp loop from the impedance variations of the incoming current.

Modern opamps have pA bias currents, or fA. There's seldom need to balance input resistances. Haven't done that in ages.

The opamp output will get digitized and go into an equivalent phase detector/lockin amp, so a little DC offset actually doesn't matter.

I'm still playing with the idea... nothing's final. Basically I want to do a 3-terminal measurement of a concentric-tube capacitive liquid level sensor. Apply a sine wave to the outer tube and measure the current induced into the inner with this sort of TIA. Long cables might be involved.

Thanks for the comments. It's good to think things over different ways.