I am not sure what are the efforts for each item, so assign efforts and starts with the cheapest one.
825 karma · joined August 11, 2017
I am not sure what are the efforts for each item, so assign efforts and starts with the cheapest one.
This might be helpful:
https://learnk8s.io/troubleshooting-deployments
I am planning to do the same for a platform that I am building, and is deployed on prem. Let me know if this is an open source project.
My question is about the operational aspect of online vs offline stores.
I.e. what happen if the features needed for a prediction are not in the ONLINE store?
To sum up, assuming that the online feature store is some sort of a cache, how do you know which objects to place in the cache?
With humans you must make sure that the same human with the same skill set who knows stats at Master Level, will always be there for your specific data and actually have the time to do the experiments.
Also, I think that 95% of the users/consumers of machine learning are non consumers - I.e. they do not have ANY access to any machine learning tech, and thus need to revert to guessing.
So the ethical thing to do is actually give them some tool even if it might not be optimal.
1) Using cross validation/validation set. 2) Regularisation . 3) Finding statistically significant features (e.g. chi square).
Why cant this be done automatically? I.e. what is the human advantage ?
I will be surprised if the salary cut includes good engineers.
Also, most of the simple DS tasks, will be automated by AutoML.
If I were you I would learn golang or Rust. You can join a data engineering group or any company that is moving to cloud native architecture.
"It does not matter if it does not sell".
So, work on it for a week, max a month, and start selling.
What should exist, is a set of standards that defines generic cloud resources, and than each cloud provider would implement a translation operator from the standard to its own CRD. Of course, the leading cloud provider has no incentive to do that since this would reduce lock in.
For example, there should be a VirtualBucket CRD which would map to AWS Bucket, Azure Container etc.
Terraform was first into this area, but terraform is still manual in nature.
I think that Pulumi uses that same code generation as Arc.
Can you give an example a landing page that you like?
As a startup you simply do not have the long term horizon for strategic projects. It would take all of your time to cover regular fires.
If you main risk is business risk (I.e. most money go to marketing/ad), then you go shallow, and make sure that everything go smooth.
If your main risk is tech risk, than you should lead the way. I.e. to lead in a tech risk startups, you must be versed in the most risky tech, other wise, you will not be able to judge the technical decision, which might lead to your ruin.
The key here is that there are no shortcuts. If your spend 4 years on one subject and related ideas, you will be better than someone who spend only 3 months.
At this point most of the model training is commodity. The game will shift to auto ml services.
Brex on the other hand looks at your company state.
I am about to release a new Kubernetes based machine learning product (using new CRDS, operators, etc).
I wonder if you can you do a paid review? we can talk offline if needed.
So to sum up, what did customers care about the most from your point of view? (maybe top 3)
However, did customers raise concerns about hosting your code in their on perm clusters, For example, did they insist that you open source?
Also, when you host the code on the customer cluster, how did you track usage? For example, was there a process within the customer site, that would send usage information to your site ?
And, I would also imagine that every engagement is unique (not sure if this is correct),
Hence, did you outsource the solutions engineering or that was done as part of the team?
Another question: can you elaborate more on how did you price the product? Did you do A/B testing on the price? ask the first customers?
What I think might work is auto ml combined with models ops, or rather auto model ops.
Or even better: auto data managmenet -> auto pre-processes -> auto ml -> auto ops.
So I am about to introduce a new product into this space, can you share how did you get your first 1-50 customers?
Also, Can you share general details about your customer demographic? I.e. what verticals, company size, etc?
I.e. you should start writing code only after you have the full design in details. In short, think of yourself as a human code generator.
The issue with programming is that you need to hold both the 3000 level picture and the 0.1 level, so working with diagrams should help.
The second thing to think about state machine, especially one that are specified by tables.
Before you write the code, make a table of each input state, and map it to your program output state. This would make sure that you did not forget anything.
Look at the work of David Parnas in this area.