Okay, the DMG2302 is worse than the ones I suggested, while the chinese clone SI2302's are cheaper (I think they are made by MicroCommercialCorp, which has cloned a bunch of Vishay's parts using the same part numbers. I think they may actually have some sort of deal with Vishay - either that, or they're just doing what China Inc. has a reputation for. Expensive from western vendors, ie, same price as Vishay's, but mysteriously they're cheap as dirt from chinese vendors on aliexpress/ebay. They seem to meet spec though... I wonder if there are royalties that aren't being paid on the ones sourced from china?)
In any event, MCC has some solid technology for low Vgs, low Rds(on), high current FETs. They've released a MCU90N02 that is just a BEAST - not ROHS though - and in DPAK package (ie, easy to hand solder - sure, the western manufacturers can beat those specs, but they do it only in packages like "POWER-TDFN8" that are hard to hand solder, and straight up cannot be reworked without mad hot air (and have fun with that if you put a properly thermal pad under it)), I'm going to be releasing boards with 4 of those on it in november... I love their systematic part naming too (on the parts that aren't clones, where they just copy the part number of the original manufacturer)
Edit: WOAH WOAH WOAH!!!!
Those are N-channel FETs. You are using them like they were P-channel FETs! No wonder you see the results you do! It's not turned on at all, because when you drive the gate low to turn it on, that turns it off, and it's conducting through the body diode. If you try to turn it off, you'll find that it doesn't turn it off!
You want a SSM3J358 or SSM3J332 or something like that... (If you don't want to solder them, I sell breakout boards, $~6 for a 6-pack with the chips mounted, and a 10k gate-source resistor)... https://www.tindie.com/products/drazzy/simple-sot-23-mosfet-six-pack-logic-level/