Data Science Architect | 9 to 14 years | Gurgaon

What matters most

• End-to-End Lifecycle Ownership

  The role requires managing the entire AI/ML value chain, from initial presales and solution architecture to production deployment and steady-state governance. You must demonstrate the ability to translate business problems into technical roadmaps while maintaining accountability for project financials and delivery quality.

• Platform Architecture and Modernization

  You will be responsible for designing greenfield cloud-native ML platforms and leading complex brownfield migrations. This requires deep expertise in selecting appropriate tooling for data lakes, feature stores, and MLOps pipelines to ensure scalability and minimal business disruption during transitions.

• Autonomous ML Operations Strategy

  A key priority is defining the roadmap for self-optimizing AI systems. This involves implementing automated retraining, drift detection, and closed-loop monitoring to move beyond manual model maintenance toward mature, autonomous operations.

• Technical Leadership and Mentorship

  As a senior architect, you must act as the single technical point of accountability, coordinating across data engineering, MLOps, and business stakeholders. You are also expected to build reusable frameworks and mentor delivery leads to standardize best practices across multi-industry engagements.

Likely challenges to think through

• Balancing Presales and Delivery

  The role demands high-level presales activity (RFP/RFI responses and commercial shaping) alongside hands-on technical governance. The challenge is maintaining technical depth and architectural integrity while managing the commercial pressures and scope constraints of multi-year transformation programs.

• Navigating Legacy-to-Target Migrations

  Leading large-scale modernization projects involves significant technical debt and organizational resistance. You must prepare to manage the trade-offs between adopting modern cloud-native architectures and ensuring business continuity for clients with entrenched legacy systems.

• Implementing Responsible AI at Scale

  Ensuring model governance, transparency, and fairness is a core requirement. The difficulty lies in operationalizing these concepts within complex, high-velocity environments where speed-to-market often conflicts with the rigorous validation required for explainable and ethical AI.

Topics to prepare

• MLOps and Autonomous Operations

  The role requires designing roadmaps for self-optimizing ML, including automated retraining and drift detection. Exercise: Draft a high-level architecture diagram for a closed-loop MLOps pipeline that integrates drift detection with automated model retraining triggers, ensuring you can explain the trade-offs between manual and automated intervention.

• Presales and Solution Architecture

  You will lead RFP/RFI responses and effort estimation. Exercise: Review a hypothetical complex business problem and outline a structured approach to solution architecture, including how you would estimate resource requirements, define project phases, and present the 'win themes' to non-technical executive stakeholders.

• Responsible AI and Model Governance

  The position emphasizes explainable AI and bias mitigation. Exercise: Prepare a framework for evaluating a model's fairness and explainability, specifically detailing how you would document model lineage and governance for a multi-year enterprise transformation program.

Skills to revise

• Generative AI and RAG Frameworks

  The role requires expertise in LLMs and Agentic AI. Exercise: Build a simple RAG pipeline using LangChain and a vector database, then document how you would scale this to an enterprise level, focusing on data privacy, latency, and retrieval accuracy.

• Big Data Platform Architecture

  You must architect cloud-native data lakes and feature stores. Exercise: Compare the architectural benefits of using Databricks versus Snowflake for a large-scale feature store implementation, focusing on how each handles data lineage and integration with MLOps pipelines.

• Program Governance and Agile/SAFe

  You will manage multi-workstream programs. Exercise: Create a sample project dashboard layout that tracks KPIs such as model accuracy, drift metrics, and financial burn rate, demonstrating how you would report progress to a steering committee.

• Prepare a case study on end-to-end MLOps architecture

  Develop a structured walkthrough of a greenfield ML platform you designed. Highlight your selection process for cloud-native tools, feature stores, and MLOps pipelines, specifically addressing how you implemented automated retraining and drift detection to achieve autonomous operations.

• Synthesize your experience with Generative AI and RAG

  Prepare to discuss a specific project where you integrated LLMs or Agentic AI frameworks. Focus on how you managed data governance, prompt engineering, and RAG implementation while ensuring the solution met Responsible AI and explainability standards.

• Draft a framework for presales and solution estimation

  Create a brief outline of your approach to responding to RFPs/RFIs. Explain how you translate complex business requirements into technical architecture, estimate effort, and present 'win themes' to executive stakeholders.

• Review program governance and delivery methodologies

  Prepare examples of how you have managed multi-year, multi-workstream AI programs using Agile/SAFe or TOGAF frameworks. Be ready to explain how you tracked KPIs, managed project financials, and mitigated risks during large-scale legacy-to-target migrations.

• Formulate strategic questions for the interviewers

  Prepare insightful questions regarding Capgemini's current approach to reusable ML frameworks, how they balance innovation with Responsible AI governance, and the typical structure of their cross-functional delivery teams.