Software Modernisation, Cloud & AI: Tarun Chatterjee
Tarun Kumar Chatterjee — Lead Software Solutions Architect, Presidio, USA
Tarun Kumar Chatterjee, a Lead Software Solutions Architect at Presidio in the USA, argues that successful software modernisation begins with the business problem rather than the technology: understand what is preventing the organisation from reaching its goals, then choose cloud, AI or integration tools only where they deliver measurable value. In this Xraised interview with host Myles, he walks through how he approaches complex client problems, what cloud transformation really offers, where AI helps and where it is hyped, and how companies with ageing systems can begin.
Tarun describes his interest as starting with curiosity about how things work and how an idea can become something practical. An early project that stayed with him was building a simulator and driver for an electronic cutting device. The software had to translate vector-based design coordinates into hexadecimal commands that the machine could execute. The experience taught him that software engineering is about understanding the whole system, including hardware, communications, user requirements and expected outcomes.
His career then moved through enterprise application development, system integration, distributed architecture and large-scale modernisation. Along the way he has worked with Microsoft.NET, Azure, AWS, APIs, microservices, messaging and event-driven platforms, and front-end frameworks such as Angular and React. What keeps him engaged, he says, is the constant evolution of the field, from service-oriented architecture and cloud platforms to generative AI and now agentic AI, and the reward of seeing a business requirement become a solution people actually use.
Tarun starts with the business challenge rather than the technology. He notes that the solution a client asks for is not always what they need. A request to migrate an application to the cloud, for example, may turn out to stem from high infrastructure costs, slow release cycles, poor integrations, scalability limits, or a hard-to-maintain legacy application.
On the business side, he looks at objectives such as cutting costs, improving customer experience, raising transaction capacity, reducing manual effort, or speeding up time to market. On the technical side, he reviews the existing architecture, application dependencies, data flows, integration points, security requirements, and performance limits. He also considers how teams work and whether the organisation can operate the proposed solution.
He breaks the problem into manageable components and validates assumptions early through small proofs of concept instead of spending a month designing a large solution on untested assumptions. He cited an enterprise modernisation involving the migration of critical jobs to a cloud-based integration approach using Kafka, supported by AI-driven process automation. The lesson he drew was that you must understand the existing environment before deciding how to improve it. He also stresses non-functional requirements such as scalability, security, monitoring, disaster recovery and maintainability, and the need to explain technical complexity in business language.
Tarun calls it a misconception that cloud transformation is simply moving applications from an on-premises data centre to a cloud platform. In his view, the real opportunity is to rethink how applications are designed, how infrastructure is managed, and how quickly the business can respond to change. He highlights several benefits:
Agility and scalability: platforms such as Azure, AWS or GCP let organisations provision resources faster and scale with demand, so they need not size infrastructure for peak load all year. Cost optimisation: cloud does not automatically cost less, and a poorly managed environment can be more expensive. Right-sizing, monitoring, and the right architecture improve utilisation. Application modernisation: moving away from tightly coupled systems towards APIs, microservices, containers, serverless computing, managed databases and event-driven platforms. Data and AI access: cloud analytics and machine learning services let teams build intelligent applications without starting from scratch. Better integration: cloud integration services and event streaming can connect applications built at different times and improve operational visibility.
His advice is to start from business objectives and define a phased roadmap that balances value against technical risk. A cloud transformation succeeds when it changes how the organisation operates, not merely where applications are hosted.
Tarun sees meaningful opportunities in intelligent automation, enterprise knowledge management, document processing, software development, customer support and business support.
He explains that a RAG-based application retrieves relevant information from an organisation's own knowledge sources rather than relying only on what a language model already knows. He described exploring an Azure OpenAI-powered chatbot using retrieval, vector embeddings, and knowledge sources to make enterprise information easier to find and to reduce repetitive manual effort. He also points to intelligent document processing, which can automate parts of extracting data from identity documents, invoices and other records.
Tools such as GitHub Copilot, Claude or Codex can assist with code generation, test creation, documentation and understanding existing code. Even so, he says AI-generated code still needs engineering judgement, with developers reviewing it for correctness, security, performance and maintainability. Agentic AI, which uses language models and tools to carry out multi-step tasks, may help with workflow automation, system monitoring, operations support and complex process orchestration.
Not every workflow needs an AI agent; some are better served by conventional automation, APIs and deterministic business rules. He lists data privacy, hallucinations, security, access control and accountability as key concerns. In his view, the future is about combining human expertise, software engineering and AI rather than replacing every application or employee.
Tarun's first recommendation is not to start by buying new technology but to understand what is blocking the business goal. He suggests a structured approach:
1. Assess the current landscape: identify business-critical applications, costly systems, integration pain points and operational risks, plus redundant or duplicate applications. 2. Prioritise by business value: not every application needs the same modernisation. Rank initiatives by measurable outcomes such as cost reduction, performance, customer experience, lower risk, and faster delivery. 3. Build a realistic, incremental roadmap: start with high-value applications or a specific process, validate the architecture, measure results, and then expand. 4. Treat cloud and AI as part of the strategy: for instance, modernising an integration platform can open the door to real-time data processing and event-driven architecture, and improving customer support may introduce RAG-based assistants where they make sense. 5. Establish governance and an operating model: modernisation does not end at go-live, so teams must monitor performance, manage cloud costs, maintain security, and keep improving the architecture.
Comments
No comments yet — be the first to share your thoughts.