I'm not an expert deckbuilder but I'll point out what I think doesn't gel in the deck.
One point to note is that influence is precious. You want to be using influence on effects which you cannot have in-faction. Also, you want to be sure that the cards brought in (aka splashed) interact with a sufficient proportion of your deck to be worth including, or have a significant enough impact that it is good to have them even if they may not appear every game. Especially for MaxX, you may end up milling the splashed cards. Hence for MaxX, either ensure that the splashed cards are important enough to have multiple copies splashed, or can be accessed or interacted with from your Heap.
My thinking was that my deck was mainly a mill deck which milled itself as much as the corp via MaxX's ability. As such, recursion cards were obviously a good choice. However, I noticed that when playing, I would often draw the big icebreakers like morningstar and Femme only to end up drawing over my limit so I could trash those cards and retrieve them using retrieval run. So I figured I'd try out "inject" as it had an interesting ability that seemed to work well with my strategy. I tried it a couple games and it works amazingly well! I play it for free using an installed voicePAD, then get to draw 4 cards to my hand. As my deck is mainly composed of events this usually gives me an instant economy boost. Plus, when I get programs instead they get trashed and I get credits which works beautifully. You're on the right track, recursion is very important for MaxX. Inject and I've Had Worse help to lessen the burden on your recursion cards by enabling you to draw a greater proportion of your deck. And as you have noticed, plenty of recursion cards mitigates the downside of Inject.
I also switched my console to Vigil as I felt I didn't need Grimoire's ability at all and that with Vigil, so long as the Corp keeps a max hand size I draw two cards every turn (once for MaxX and again for vigil) which is like Wyldeside. Plus it gives me an MU. Vigil costs 1 credit less than Grimoire, but also gives you 1 MU less. That may not be a problem for some decks but if you are splashing for Akamatsu Mem Chips, then it might be for your deck. Getting +2 MU with your current arrangement requires 3 credits, 2 clicks and drawing both cards to install both hardware. With Grimoire, you save 1 click and 1 draw. You are probably better off saving the 2 influence on the Akamatsus and having multiple copies of Grimoire instead of 1 Vigil + 2 Akamatsus.
Secondly, Vigil's ability is less useful for MaxX cause it has lesser opportunities to fire over the whole game. You start with 5 cards in your Grip, with 40 cards in the Stack. With MaxX's ability, Injects and IHWs, Vigil has a limited number of turns where it might be able to fire. This number becomes even more limited if you consider that the Corp can manage their hand size to deny you the free draw and also that the single copy of Vigil might get trashed and it is definitely not a card worth recurring.
Grimoire is not that useful for you because you have no viruses in your deck. However, you dropped Parasite which is one of the most powerful cards Anarchs have. Think of Parasite as a run economy card. At some point, it trashes the ice it's hosted on so you no longer have to break that ice whenever you run that server. If that ice trashed was say Ichi 1.0, you just saved yourself up to 5 credits for *every* run on that server. It also reduces the ice's strength along the way so even before then it is saving you credits on pumping your breakers (or it puts the ice in range of Morningstar). I think you should consider putting Parasites back in, especially considering you have 3 Deja Vus in the deck.
Morningstar was nice but once people rez a curtain wall or even a fire wall advanced 1, it's useless. Sure, Femme can bypass it but when playing against Blue Sun they'll just bounce the ice and remove the mark. I wanted to be able to break any ice but still maintain my low program to event ratio for other cards like inject. I finally settled on: Keyhole + Eater for the R&D mill. D4v1d for anything 5 strength or over (it's in faction and I added e3 implants to prolong it). Morning Star for the 5 strength or lower barriers. Cerberus "Cuj.0" H3 for the 4 stregth or lower sentries. Overmind for anything inbetween. With your current breaker suite, if the Corp plays a Swordsman on R&D you cannot run R&D until your single copy of Cuj.0 is drawn. Despite your massive draw this is a huge problem as it could delay your Keyhole runs by several turns, during which the Corp has plenty of time to make R&D excessively taxing to run on. The original decklist had 5 answers: 2 Femmes and 3 Parasites. Also without Femme and Parasites, a Crisium Grid on R&D effectively locks you out permanently.
If your range of sentries targeted by Cuj.0 is strength 4 or lower then the break cost difference is not that big especially if you factor in the limited usage of Cuj.0. Femme also has the additional bypass ability which can save you a lot of credits and can be installed for cheap with Retrieval Run.
The thinking with overmind was that, early game I install Eater + keyhole then mill for all it's worth. once I'm satisfied, I install Overmind, trashing the Eater + Keyhole combo, then run Archives for the win with 3 counters on overmind and e3 implants to keep it going. My console already aidds to it's appeal by increasing my MU. I originally put in Q-coherence chip since it cost nothing to install but realized it doesn't work with what I want to do since it gets trashed whenever a program is trashed, even by mine own hand! So instead I added Akamatsu mem chip for a cheap MU boost. Ideally, I would have both chips out plus vigil bringing my total MU up to 7 then installing overmind with 6 power counters on it. I was worried this might be too slow but it actually happened my last match versus Blue Sun.Overmind is too narrow in its purpose for your deck and too costly. 2 deck slots and having to trash all your other programs to use it is way ineffective. Consider swapping in Crypsis instead. Only 1 credit more to install but you can click to increase the number of ice it can break. With Grimoire installed, it even comes with one free token!
A single E3 is also likely as not to be trashed in a MaxX deck. Plus it only interacts with 3 out of 45 cards. You can probably spend the influence on something better like the third copy of Lucky Find because each copy is a major econ boost in a PPVP deck. Also the third copy of Sure Gamble is also something you should be keeping not only because it is such a strong card to have in the starting hand but also because it is another great econ boost in a PPVP deck. Day Job on the other hand is something I feel iffy about for MaxX cause I am not sure if MaxX can afford that kind of downtime. I would rather cut one copy of Day Job than Sure Gamble or Lucky Find.
Hmmm, what else?Scavenge is nice to have but requires that you have something installed which you no longer want. You have 5 cards which you are happy to Scavenge, out of a total of 11 programs. That does not give you much to Scavenge. Do consider Clone Chip instead because it gives you the ability to instantly install the program you need mid-run which is something you cannot get in-faction. Comparatively, Scavenge vs Uninstall or Deja Vu is a savings of the install cost and a click which arguably is less important for this deck.
I added Same Old Thing to make sure I get that Levy AR when I need it. I added Scavenge to make D4v1d and Overmind even more appealing since, as I understand it, once they run out of power counters I can use Scavenge to trash the useless card and reinstall that exact same card for free. Since I did put e3 feedback implants in Cuj.0 made more sense to me (I was previously using Alias).
-1 Day Job
-2 Akamatsu Mem Chip
-1 E3 Feedback Implants
-1 Cerberus "Cuj.0" H3
+1 Sure Gamble
+1 Lucky Find
+2 Clone Chip
+2 Femme Fatale
(In my opinion, having a few extra cards in a MaxX deck does not matter much. But if one insists on keeping to the minimum deck size, the deck would need more testing before one knows what to cut.)