With that said, if you really have the skills and time, then the best service you can offer would be miners repair, if you build the right team for it, you will make a lot of money, the only thing that is lacking in supply in all this mess are companies that can fix mining gears.
The problem with repairing mining equipment is it's difficult to find enough electronic repair technicians who do it at a large scale. Repairing hashboards is probably a skill that takes months to learn, not accounting for the dozens of ASIC models on the market. BGA rework is expensive, so it's probably cheaper just to buy a new hashboard.
What kind of software are you planning on writing? the mining field is pretty limited in that aspect, it's either a mining pool, a monitoring software, or firmware, none of these require having a large farm, just a few gears for testing, and a ton of coding skills and free time.
I do think there are opportunities for things like devfee removal, automatic shutoff when the electricity cost is too high (large farms usually have realtime energy prices) and even a web application for managing the customer relation stuff.
You're right that I don't need a large farm to do this kind of development. I can only afford a small warehouse with 50 ASICs / 200 GPUs (after ETH PoS) anyway. But I have no better investment to put the money into, and I think showing the medium-sized farm to potential customers will help sell the software and consulting. No other investment has the same tax benefits that mining does except maybe real estate.