How do I approach this seq design?

I don’t understand the description in the picture Maybe you could describe an example signal flow?

Or just a brief description what the general functionality of the sequencer is and what the knobs do?

Just saw post! Will explain... just a sec...  edit: ok...

1. A count is produced from the clock, the sequence length being determined by the "Outer cycle" OC parameters 1-64,96,128,192 & 256. Within that count loop, an "Inner cycle" IC with the same parameters. The IC is a loop within the OC and is modulo of the OC; mod(nfromOC,IC), So the count would be; 0 to IC to 0 and count until the OC count has finished, then repeat. The pendulum is determined by the IC length. When the count reaches just over half way, the count is reversed (mathematically) for the other half of the count; If nfromOC>((IC*.5)+.5) then; ICn=(IC-nfromOC) else; nfromOC. Whatever the count, it is wrapped around the 8 CV pots.
The count/shift adders, are the same, should have given them the same name!... Three of them would be better. One before the OC moves the whole sequence phrase position, the one in between changes the flavour or order of the pattern and one on the end would move the sequence up or down the 8 pots, all of them useful.

2. The resultant count which drives the sequencer, is transformed thru a "complex" lfo to extract different but related, to a 0-5V CV pattern which is then used to drive the gate comparator. 

3. The Gates are: High, Gate and pulse, the pulse being the shortest gate, .001 to .03 secs.
Not all gates change the CV pots tho, 2 modes for a S&H gate for pot changes would be: mode1; S&H gate on a high, if two notes played consecutively, just the initial high. mode2; S&H gate on "gate" as well as on initial"high".
The comparator works like this. The top pot selects a portion of the -5V lfo CV, it has selection priority over pots underneath. If turned all the way up, it would steal all of the modulation and play a continuous high with just the initial pitch CV selected with the S&H gate. The higher pots have priority over the ones underneath.

Although there are only 8 pots, sequences from the comparator will pick up on different CV notes on long sequences. I've tried this out in audulus and it seems to glue together. I'd be open to a different complex LFO method to extract patterns from the sequencer action but the method I used does have a good varied result. Multiplying down the sequencer count and putting it through a sin function brings dramatic variations, doing the same but through a fract/modulo function gives variations of the extreme transform, and phase helps to shift the pattern along predictably. Varying pots on the gate comparator with the lfo during a sequence produces a lot of related variation. 

...hope I explained that ok. ...edit: except not so brief, sorry. 

Okay, sorry, still lost, BUT it sounds crazy and that’s awesome! From the little bits I understand, it seems that it should be a piece of cake for the Arduino as long as you know how to code. You just have to see about the analog CV output, which either requires PWM with filtering (which isn’t too difficult and might just require a little trial and error with the R-C circuitry) or an external solution (e.g. DAC IC).

Reading analog CV from the AE Modular, as well as reading and outputting gates is trivial though

PWM will be fine I think if you find the right resistor-capacitor ratio. The most precise solution is an external DAC IC, controlled by the Arduino.