UAE
Pushker Sahai
Sovereign AI & Data Platform Leader
Google (San Francisco) · Microsoft (Seattle) · Cloudera (Toronto)
I help enterprises design and deliver governed AI and modern data platforms—secure, audit-ready, and scalable across cloud and hybrid environments.
What I do
Sovereign AI & Governance
Data residency, access control, model lifecycle governance, auditability, and Responsible AI aligned to regulated environments.
Modern Data Platforms
Warehouse/lake/lakehouse architectures with lineage, quality, and operating models that unlock trusted analytics and AI.
AI Engineering at Scale
Production RAG + multi-agent orchestration, LLM evaluation/observability, LLMOps, and performance/cost optimization.
Featured work
CTO: Multi-Agent Orchestration Platform
Led architecture and execution for a multi-agent orchestration layer built for reliability and horizontal scale (designed for 100K+ concurrent users).
Enterprise AI & Data Reference Architectures (Google)
Drove executive workshops and reference architectures for governed AI and modern analytics—accelerating POV → production with security and standards.
Enterprise Modernization & Delivery (Microsoft)
Directed modernization roadmaps and delivery execution across cloud/hybrid estates—aligning architecture, operating model, and compliance for scale.
Hybrid Architecture & Governance (Cloudera)
Designed regulated-ready warehouse/lake/lakehouse options and guided POC/POV lifecycles to production adoption with governance and lineage.
How I help
Sovereign AI Readiness (2–3 weeks)
Assess target use cases, data residency constraints, and governance gaps. Deliver a pragmatic control framework and reference architecture.
- •Use-case shortlist
- •Governance controls
- •Reference architecture
- •Implementation plan
Data Platform Modernization Blueprint (3–6 weeks)
Define target architecture and migration waves for a governed warehouse/lake/lakehouse foundation that supports AI at scale.
- •Target architecture
- •Migration waves
- •Operating model
- •Success metrics
AI to Production Accelerator (4–8 weeks)
Stand up production patterns for RAG/agents with evaluation, observability, security guardrails, and deployment workflows.
- •LLMOps patterns
- •Eval + observability
- •Security guardrails
- •Pilot-ready system