Most US government agencies don't have discretionary money to spend. That means that you can have the greatest idea in the world, and still may not be able to find a buyer.
Selling to the government usually entails one of three things:
1) You can sell a commodity by being a registered vendor, e.g. through the GSA. GSA registered vendors have usually agreed to prices up front, which eliminates some of the red tape. But this doesn't seem to match your situation.
2) You can find an announced opportunity. In other words, the government is telling you that it wants something. You can then write a proposal to sell it that thing. These opportunities are normally advertised in FedBizOps. They are openly competed. The catch here is that you have to be able to provide what the government tells you it wants. You can't tell it what it should buy.
3) You can respond to a SBIR, BAA, or other R&D award announcement, for instance from NASA, the NIH, the CDC, etc. This seems to fit your situation most closely. However, as with #2, these announcements are generally tailored; they are intended to produce a final product or prototype, which performs some function that the agency already wants. If they don't want what you have, you're out of luck. But unlike FedBizOps announcements, they are often more open-ended, and hence more suited to new ideas or technologies.
You can also play the long game: develop some neat tech, and then try to build support for it within government, which then leads to an opportunity like #2 or #3. This entails a lot of behind-the-scenes politicking, which is what (for instance) large defense contractors are good at. It may entail lobbying Congress, or getting your tech in front of generals, admirals, or senior agency management, and then being able to wait a few years until the next appropriation cycle (or two or three) rolls around. For obvious reasons, this is difficult for startups and small businesses to pull off, especially if you don't already have personal connections to the agency you are selling to. Successful small businesses that sell to government are often started by people who have these ties already in place when they start.
If you see opportunities on FedBizOps where you only have a partial solution, look to see which Govt contractors/suppliers are already working/bidding in that area. Often you can sell your partial solution to them first, and they package that into a larger solution that is sold in to the government opportunity. This is still a complicated long term sale but it can get you on the board.
Once you have smaller wins at this level, leverage them to drive revenue, product feedback and customer success stories for related conferences/marketing materials. To promote success stories - look at government focused media such as e-republic's various brands. (http://www.erepublic.com/) or specialised media (there are a bundle focused on health/defense etc). Check their editorial calendar and network at their or related industry events.
Bonus points for landing government contracts if you can fit into one of the underrepresented/minority/disadvantaged group or small business categories.
There's basically a quota that some percentage of contracts must be awarded to those groups.
-Make sure you qualify as a “small business” under SBA regulations
-Make sure you have the necessary accounts (DUNS number & CAGE code) to accept Purchase Orders and be awarded a government contract.
https://www.sam.gov/portal/SAM/#1#1
-For instances in which the agency may pay you via credit card, make sure that you have some way of accepting credit card payments that does NOT use a third-party intermediary like PayPal.
This is one of the reasons why you will see Alaska native corporations winning US DoD/fedbizopps contracts. They essentially act as a GC and then implement the project using eight different specialized subcontractors for all of the unique tasks.
Items 1, 2 and 3 pretty neatly cover federal government level stuff. At the more local level (City, County and State), the procurement/purchasing departments of each organization generally have a process for posting RFP/RFQ for specific projects. Basically the same idea as fedbizopps postings but on a smaller more local scale.
If you want to bid on things like this at the mid-sized-city or county scale, it's usually a process of first being set up as a legal business entity in the jurisdiction, then making sure you meet all of the other eligibility requirements, and then contacting their purchasing/contracting departments to register yourselves as a vendor and actually participate in their RFP/RFQs.
But I am not sure how much any of that would apply to a ML/healthcare thing where you want to propose new novel solutions/technologies which are not explicitly called for in an RFP/RFQ. State projects tend to be much more like "we are looking for a general contractor to erect a 150 ft radio tower on this mountaintop as part of the highway patrol/emergency communications radio network upgrade project", or "we are looking for a contractor to build some giant concrete half pipes over top of I-90 and then cover them with dirt+landscaping to create a wildlife overpass". Where they already have a very clear idea of what they want before it gets to the RFQ stage.
I’ve been PI on a number of SBIR and other us government funded R&D projects. In my experience, the way to win an award is to market the idea to a potential govt sponsor. Sell them on your solution and then try to get them to write a public solicitation the general specs and requirements that exactly match what you have/want to do. It doesn’t guarantee funding, but is way higher chance than just responding to a solicitation you just happened upon.
Selling to the government usually entails one of three things:
1) You can sell a commodity by being a registered vendor, e.g. through the GSA. GSA registered vendors have usually agreed to prices up front, which eliminates some of the red tape. But this doesn't seem to match your situation.
2) You can find an announced opportunity. In other words, the government is telling you that it wants something. You can then write a proposal to sell it that thing. These opportunities are normally advertised in FedBizOps. They are openly competed. The catch here is that you have to be able to provide what the government tells you it wants. You can't tell it what it should buy.
3) You can respond to a SBIR, BAA, or other R&D award announcement, for instance from NASA, the NIH, the CDC, etc. This seems to fit your situation most closely. However, as with #2, these announcements are generally tailored; they are intended to produce a final product or prototype, which performs some function that the agency already wants. If they don't want what you have, you're out of luck. But unlike FedBizOps announcements, they are often more open-ended, and hence more suited to new ideas or technologies.
You can also play the long game: develop some neat tech, and then try to build support for it within government, which then leads to an opportunity like #2 or #3. This entails a lot of behind-the-scenes politicking, which is what (for instance) large defense contractors are good at. It may entail lobbying Congress, or getting your tech in front of generals, admirals, or senior agency management, and then being able to wait a few years until the next appropriation cycle (or two or three) rolls around. For obvious reasons, this is difficult for startups and small businesses to pull off, especially if you don't already have personal connections to the agency you are selling to. Successful small businesses that sell to government are often started by people who have these ties already in place when they start.