They'll need to grow in pods
Since you're doing genetic engineering anyway, have each insect grow in its own little cone-shaped pod. Even if dropping from a point that is stationary relative to the atmosphere, I think your insects will reach hypersonic velocities before aerodynamic drag is high enough to slow them down (see some math below, and the transitional drag paper below for more math and some pretty graphs). They will still experience heating, and may tumble hard without some aerodynamic help. They're also going to have to deal with subzero temps after they do slow down, which an insulating pod will help with. You could optionally make the pod vacuum-tight as well, with a little vent or trapdoor that seals from the inside with some sort of secretion. The little pod can be ablative -- it turns out that the material emitted from ablation helps slow down at these sizes (also in the paper).
Hypersonic Housefly
As I mention above, deployment speed may not matter much. The potential energy even at a "stationary" point above Earth's gravity well is impressive; all that potential energy will get converted to kinetic on the way down.
Play with calctool, referencing upper atmosphere models and various drag coefficients to get a better feel for this. For example, calctool will tell you that the terminal velocity of a 10 mg housefly at 100 km altitude is 6213.60 m/s (assuming a drag coefficient of .5 and a cross-sectional area of around 20 mm^2).
But, as @JanHudec points out below, our fly won't reach terminal velocity at 100 km (but will hit terminal at slightly lower altitudes). The speed it will reach at 100 km depends on how high it's dropped from. If we somehow arrange for a stationary release at 600km altitude, then gravity will still have the fly screaming straight down at close to 3000 m/s as it enters the upper atmosphere at around 100 km.
The NASA Glenn Mach number calculator tops out at 76 km, but it tells us that 3000 m/s at that altitude is just above Mach 11. At hypersonic speeds, the fly will experience compressive heating from its own shock wave. The GRC stagnation temperature calculator also tops out at 76 km, but should help with heating estimates.
For the several seconds it takes to slow down, this fly's temperature is going to be somewhere above the boiling point of steel. The transitional drag paper tells us that something the size of a housefly will slow down sooner if ablating though, which again puts it back into a pod.
By the time our fly is down around 80 km, after dumping all that energy into heat, it's slowed to the terminal velocity for that altitude of around 1000 m/s, or about Mach 4. It looks like the fly finally goes subsonic and starts cooling off around 50 km.
At 20 km altitude, our fly will finally be drifting down at a lazy 14.4 m/s, but now has the opposite problem, because at that altitude the ambient temperature is around -67 C. After a few seconds of blistering heat, it's now going to spend maybe half an hour drifting down at subzero temps.
LEO orbit is typically around 7000 m/s anyway, so in summary you may not be gaining as much as you'd think by dropping from an atmo-stationary point. Because insulation matters more than deployment speed, with a pod you could go back to dropping from a normal low orbit so you can get better swarm distribution, which dropping from a stationary point won't let you do.
Further Research
For more background on speeds and conditions at higher altitudes, look for papers like The Transitional Aerodynamic Drag of Meteorites -- that paper, for instance, covers micrometeorites (particle sizes of a few microns), as well as mesometeorites (particle sizes of a few millimeters). Insects are going to perform somewhere in that size range.
For lower altitudes and speeds, take a look at jumps like Felix Baumgartner's, and subspace balloon flights like those of JP Aerospace -- they've learned a lot about the environment around 100,000 feet, but you'll need to get down to that altitude first.
