Head - Data & AI CoE

1. CoE & Operating Model:

- Design and operationalize the CoE: charter, federated model (central platform + BU pods), talent architecture, multiyear roadmap across all entities.

- Own the Data & AI product portfolio: prioritized by business impact x feasibility, tracked with documented ROI for each BU.

- Establish CoE KPIs: models in production, time-to-insight, cost avoidance, revenue impact, AI literacy index, and data trust score.

- Introduce data product thinking: treat datasets as products with owners, SLAs, consumers, and version control, not as outputs of one-off projects.

2. Data Platform & Architecture:

- Group-level data Lakehouse: real-time OT ingestion from 42-plant MES/SCADA/PLCs; unified data mesh across BU domains.

- SAP S/4HANA MDM, metadata cataloguing, data quality governance across all entities.

- MLOps / LLMOps infrastructure: model registry, CI/CD for ML, drift monitoring, RAG architecture standards, hallucination guardrails for manufacturing-critical applications.

- GenAI architecture decision: Azure OpenAI vs. open-weight LLMs (Llama/Mistral) vs. on-premise for IP-sensitive manufacturing data.

- Self-serve analytics enablement: build the platform and semantic layer that allows BU business users to answer their own data questions without engineering dependency.

3. Data Observability & Data Trust:

- Own data observability as a first-class function not a monitoring afterthought. Establish Lumax's data observability programme covering the five pillars: freshness, volume, schema, distribution, and lineage across all pipelines and BU data products.

- Define and enforce data SLAs and data contracts between producers and consumers; any dataset serving a plant decision or board dashboard must carry an explicit data contract.

- Implement tooling for automated anomaly detection, root cause analysis, and mean-time-to-detect (MTTD) and mean-time-to-resolve (MTTR) measurement across data pipelines targeting industry-leading MTTD of minutes, not days.

- Build a data trust score framework: every data asset in the Lumax ecosystem is assigned a trust score visible to all consumers, covering completeness, freshness, lineage depth, and SLA adherence.

- Govern data lineage end-to-end: from MES sensor at plant floor to GMD dashboard; every transformation, join, and aggregation documented and auditable; critical for DPDP Act compliance and OEM/JV audit readiness.

- Evaluate and recommend best-fit observability tooling: Monte Carlo.

4. MLOps, LLMOps and AI Engineering:

- End-to-end MLOps: feature stores, model versioning (MLflow), A/B testing, production SLAs by risk tier (plantsafety models: Tier 1 highest rigor; internal productivity tools: Tier 4 lightweight).

- LLMOps: prompt governance, RAG standards, hallucination monitoring, human-in-the-loop for manufacturing-critical model outputs.

- Model cards and AI audit trails for every production model covering training data provenance, performance benchmarks, known failure modes, and approved use contexts.

5. Data Governance and Responsible AI:

- Enterprise data governance framework: ownership, stewardship, lineage, data contracts, and data quality SLAs across all entities.

- AI Ethics Charter: DPDP Act (India) + GDPR (JV data flows) compliance; AI risk tier classification Tier 1 - 4.

- Data access control and privacy engineering: column-level security, dynamic data masking for PII, and purpose-bound data access aligned to DPDP Act 2023.

- Experimentation culture: institutionalize fast-fail as learning; documented retrospectives, cross-BU re-use of findings; psychological safety is a formal leadership KPI, not a cultural aspiration.

6. Talent, culture & AI literacy:

- Build CoE team: Data Scientists, ML/AI Engineers, Data Engineers, MLOps Engineers, Data Observability Engineers, Analytics Translators, BI Developers + BU data liaisons.

- Lumax Intelligence Brief: AI literacy roadmap for 10,000+ employees, differentiated by persona (plant operator - C-suite).

- CoE talent brand: MNC/Big Tech lateral hires; structured data career pathways.

- Establish data engineering culture benchmarks adapted to Lumax's manufacturing-first context.

7. Stakeholder Partnership and Value:

- Trusted partner to all BU CEOs, 42 plant heads, 10 JV leads. Every AI initiative carries a business case and benefits realisation plan per BU.

- Automotive value chain intelligence: build data products that serve not just Lumax's internal operations but generate insights across the OEM-Tier1-supplier value chain including OEM demand signals (Maruti, Honda, Tata, BMW India supply chain), JV performance analytics, and aftermarket channel intelligence.

- OEM relationship data advantage: own the data strategy that makes Lumax the most analytically transparent and predictive Tier-1 partner for OEM customers.

- External representation: industry bodies, automotive AI councils, CII/ACMA data forums, Tier-1 thought leadership.

Skills

Data Science, Artificial Intelligence, Data Analytics, Data Governance, Data Management, Digital Transformation, AI Transformation, Analytics Head, Analytics