Data Architecture for your next AI use Case
Thanks for checking out my article Build AI Ready Data Infrastructure 🌐✨
This article is focused on, How as a CxO or Head of Data and AI, can build an AI ready data infrastructure for your organization.
For more insights on Digital Transformation, AI, and Industry 4.0, simply hit the 'Follow' button. Follow me on Linkedin and Twitter. 🚀
Generative AI has taken the corporate world by storm, and many more organization have jumped in on leveraging Generative AI or AI for that sake to make business decisions.
According to Venture bit, over 70% of the organizations are experimenting with Generative AI.
Credits: Venture Bit
While this is really great stat, majority of the organisations are feeling they are not yet ready for Generative AI Applications. While Generative AI efficiently handles unstructured queries and data, to train the model, it needs lots of structured high quality data.
- Is Your Company’s Data Ready for Generative AI?, HBR
The Data Infrastructure and strategy for majority of the organisations is not yet ready for serve conventional AI use cases, and Gen AI needs even higher standards and higher amounts of data for training the models.
Through this article let's explore the simplest approach for building an AI Ready Data strategy and architecture.
The AI ready data strategy will always be an evolving live document, which gets updated as you have new use cases. Overall at strategy level you will be putting few key infrastructure pieces, however the architecture keeps evolving depending on -
If you need inferencing realtime vs batch
Depending on the size of data needed for training purpose.
complexity and availability of the data
Data Quality
Hence the best approach is to start your Data strategy for AI use cases is to start with a specific use case/s in mind. Here are the few typical use case most of the organisations are going after -
Enterprise chatbot - (GenAI)
Fraud detection
Search- Internal- Enterprise information system or knowledge base to leverage internal knowledge - (GenAI)
Churn Predictions
Propensity models
Credit Underwriting
LTV predictions
Here are customer success stories from OpenAI for your reference.
Once you have your potential use cases, identify which datapoints, and which data sources do you need. You will also need to know if you want to predict realtime or non-realtime, depending on this the solution architecture will differ a lot.
Prepare a list of Data sources and potential datapoints. You should consider all potential datapoints to be included.
Understand the data in terms of volume, velocity and variety.
Here is a sample data architecture in this case.
Sample block diagram for AI model training framework- TransformTechX
This is one of the key considerations for AI ready data. Data Quality can be defined by a simple acronym- CACTUS
Completeness
Accuracy
Consistency
Timeliness
Uniqueness
Sorted
Refer following post-
https://www.linkedin.com/feed/update/urn:li:activity:7216977005081190400
Data Governance is always hype as a a department which STOPS all the development and a roadblock. The right way to think of Data governance is how to use the data more offensively to grow the organization or revenues without being in trouble.
Check out following article on Data Governance - What is data governance and why does it matter?
We will keep the discussion on Data Governance some other time.
Is Your Company’s Data Ready for Generative AI?- Harvard Business Review
6 Generative AI Use Cases (2024): Real-World Industry Solutions- eWeek
The AI Readiness Gap: Why 90% of Companies Are Falling Behind by Reddy Mallidi
Join Amit on Peerlist!
Join amazing folks like Amit and thousands of other people in tech.
Create ProfileJoin with Amit’s personal invite link.
0
6
0