Engineering suite of "Intelligent Applications" for the enterprises

In the last few years, we have witnessed the proliferation of AI getting adopted in the enterprises to solve some of the critical problems. This is true and applicable across all the major industries - including - Finance, Insurance, Telecom, Healthcare, Entergy, Internet Services and the like. The most apparent outcome of this wave of change is that the traditional applications are now being disrupted and are getting replaced with "Intelligent Applications".

In this talk, we would explore what makes "intelligent Applications" different from the traditional application and what are the technical challenges in designing, architecting, deploying and maintaining such systems at the enterprise scale. The talk would cover some of the important details in designing the enterprise-wide suite of intelligent applications, some of the architectural considerations and design patterns, and the best practices.

2 favorite thumb_down thumb_up 4 comments visibility_off  Remove from Watchlist visibility  Add to Watchlist

Outline/Structure of the Talk

Welcome Remarks

Define intelligent applications

How are they different from traditional non-intelligent applications

Technical Challenges - What it takes to bring intelligence to applications

Architectural Considerations and Design Patterns.

Best Practices on designing, deploying and maintaining intelligent applications.

Tools, Technologies, and Frameworks that are emerged in the last few years to tackle these challenges


Closing Remarks

Learning Outcome

Understanding of what Intelligent Applications are, how best to design, build, deploy and manage them.

Target Audience

AI Engineers, Architects, and Data Scientists.

Prerequisites for Attendees

Prior knowledge in building products and exposure to machine learning is desired.

Prior knowledge of operationalizing machine learning systems would be good to have.

schedule Submitted 6 months ago

Public Feedback

comment Suggest improvements to the Speaker
  • Naresh Jain
    By Naresh Jain  ~  3 months ago
    reply Reply

    Hi Rahul,

    Thank you for the proposal.

    Can you please take a specific usecase and provide details around what kind of architectural considerations you had to put in place and also what specific design patterns you had to use?

    • Rahul Tanwani
      By Rahul Tanwani  ~  3 months ago
      reply Reply

      Hi Naresh,

      Thanks for your message. Firstly, I would talk about the challenges that are unique to the intelligent application (as compared to traditional applications) and how could those be addressed through architectural choices. Among other things, I would cover:

      • The role and importance of encapsulating machine learning services into a separate ML Platform (isolating intelligence (ML) from applications)
      • The architectural choice to productionize self-learning system (synchronization between learning and serving infra)
      • Various models of deployment (serving) and the corresponding trade-offs.
      • [Stretch] The role and importance of feature stores. 

      I would be happy to elaborate on any of the items above if required. 

      • Naresh Jain
        By Naresh Jain  ~  3 months ago
        reply Reply

        Thanks, Rahul. What about design patterns?

        • Rahul Tanwani
          By Rahul Tanwani  ~  3 months ago
          reply Reply

          Hi Naresh,

          Design Patterns (or rather abstractions if I may call) would be discussed throughout the talk. Few areas where they become important:

          Necessary abstractions in ML platform to build a suite of intelligent applications on top.

          Abstractions and Encapsulations for building various model interpretability/explainability techniques. 

          Various patterns of model deployments (batch systems, real-time/streaming systems, online systems, separate inference service vs embedded in the JVM ).

          Would be happy to elaborate more on any of the items.