NashTech Insights

Benchmark your ML projects

Ram Indukuri
Ram Indukuri
Table of Contents
blockchain, people, shaking hands-2850276.jpg

Over the last few years, many forward looking organizations deployed Machine Learning applications in their businesses. However, AI/ML technology choices at the time were not mature. This lead to poor quality of implementations resulting in myriad of issues.


Following are some common issues encountered by organizations.

Organizations tend to ignore these issues due to boiling frog phenomena. To objectively judge your quality of ML applications, Knoldus has developed the following dead simple frame work.


Click here to access the sheet and assess yourself. Use self rating column to enter values ranging from 1 to 5. 1 being the worst performance and 5 being the best. Make sure you are unbiased.


Hopefully, it did not take long to allocate points. Lets discuss the results.


 You have a problem. Glad you admitted it. Now thats half the job done. Its probably time for you to leap frog into new generation of ML infrastructure provided by vendors like google and achieve your AI goals by standing on top of their shoulders.

You could potentially re-engineer, provided the team that implemented this is high caliber engineering team with operations focus. If not, you should look at re-implementation. Knoldus can help with assessing and suggesting right way to go about and cost of doing it. We can also help clarify the cost vs benefit analysis.


Congratulations!. This either means, you implemented the application after MLOps matured, or you had an awesome team and did a great job of Ops focused deployment, even though the tools are not mature. Thats a commendable feat!

What we suggest in that case is to take a look at modern AI platforms, like Vertex AI powered by Kubeflow and re-engineer your existing system to the new platform. This should cost you relatively less, since you implemented most best practices anyways.

Ofcourse, there is still a large spectrum here, ie, project close to 50% vs 80%. But still its worth engaging a competant MLOps architect to take a look.


Kudos !! We consider you as a mature AI organization that is a leader. Your AI platform supports fast rollouts of new AI programs and run them with operational maturity. If you are not using AI widespread across organization, talk to us about use cases for AI that can push your business to be the leader in the industry. Most likely, you are already there. We would love to learn from your journey.


Please leave comments below and suggest if we missed any criteria. Hope this exercise helped you in understanding the maturity of your AI program.

Ram Indukuri

Ram Indukuri

Head of AI & Advisory Consultant

Leave a Comment

Your email address will not be published. Required fields are marked *

Suggested Article

%d bloggers like this: