from Hacker News

Ask HN: How do I find ML consulting clients?

by rontoes on 4/30/20, 4:25 AM with 1 comments

I want to make the move from in-house machine learning work to working as a consultant. I have a good grasp of how much I should charge, but I'm at a loss for how to find clients. My expertise is in medical computer vision but I have experience working with many other data types. Should I limit myself to one small niche?

What is a good way to start finding clients? I have attempted to make a list of potential companies from LinkedIn, but it also occurs to me that companies I find through LinkedIn are more likely to already have a sizeable ML/AI team. That in turn means it is less likely they have a need to contract some of their ML work.

by brudgers on 4/30/20, 4:09 PM
A company with an inhouse team of ML/AI workers is the best candidate for consulting because it already values ML/AI and has a substantial budget but limited capabilities (because existing staff doesn't know everything (and neither do you)). In addition, those organizations contain people in the industry with whom to network. Person A at company 1 used to work with Person B at company 2 and Person B has moved on to company 3.
Even more important the worst thing you can do is give yourself excuses to avoid the possibility of rejection. Your choices are either no work (unless there's a miracle) or facing the possibility of rejection.
Consulting is a long term business. Relationships matter. Nearly every potential client won't have work right now. The best clients already have consultants. Consulting is about digging up work more than it is about doing the work...you can always hire people less capable of digging up work for that.
A contemporary idea in sales is to maximize the number of rejections. Cold call for ten or twenty rejections a day. Not one success per month.