Principal AI Architect.

About the organisation.

Our client is a leading global technology company in the computer software industry, delivering data services, subject matter expertise, and technology solutions across multiple domains. Its work spans data analytics and AI solutions, AI-powered operations, and education and learning, serving business information providers worldwide. The organization combines advanced technology with human experts in the loop to build data assets that power products and workflows for clients across 30 countries.

About the role.

The Principal AI Architect serves as the chief technical authority for next-generation Generative AI and autonomous agent initiatives in high-throughput, heavily regulated enterprise environments. The role involves analyzing ambiguous, mission-critical business challenges and deciding which architecture is required, whether single-agent co-pilots, hierarchical supervisor-worker topologies, or choreographed multi-agent networks. The architect builds these systems on scalable Google Cloud Platform infrastructure, powered by frontier reasoning models, while meeting rigorous regulatory compliance and low-latency requirements.

What you will do.

• Analyze complex business friction points, evaluate trade-offs between reasoning accuracy, latency and cost, and determine the optimal architectural paradigm.
• Establish model selection frameworks across commercial frontier models and design intelligent routing gateways based on task complexity, inference cost and SLA thresholds.
• Define enterprise-wide blueprints, design patterns and engineering standards for agent state management, memory persistence and inter-agent communication.
• Evaluate orchestration frameworks such as LangGraph, LangChain, AutoGen and LlamaIndex against lightweight native implementations to meet strict latency SLAs.
• Implement agent-tool interoperability standards such as Model Context Protocol to separate model reasoning from backend transactional execution.
• Architect real-time, read-only federated data access layers across heterogeneous enterprise platforms without duplicating data into monolithic data lakes.
• Design high-throughput, low-latency pipelines, asynchronous event streaming and in-memory feature caches to meet sub-second and sub-50ms execution targets.
• Architect policy-enforced access control, compliance safeguards, PII redaction, prompt injection defense, audit logging and customer-managed encryption key management within enterprise perimeters.
• Design calibrated autonomy frameworks that progress workflows from co-pilot recommendations to human-approved actions and bounded autonomous execution.
• Architect automated evaluation frameworks, golden test datasets and shadow-execution pipelines to validate reasoning, mitigate hallucination and benchmark accuracy before rollout.

What you bring.

• 8+ years in distributed systems architecture, enterprise software engineering and data platforms, with 2+ years architecting and scaling production Generative AI or autonomous agent systems.
• Proven ability to dissect ambiguous operational problems, formulate architectural alternatives and rigorously defend trade-offs.
• Deep technical mastery of frontier reasoning models such as Anthropic Claude and Google Gemini, including prompt caching, function calling, XML structuring and token optimization.
• Hands-on experience with multi-agent orchestration frameworks such as LangGraph and AutoGen, and integration standards such as Model Context Protocol.
• Extensive experience designing on Google Cloud Platform (Vertex AI, Cloud Run, Pub/Sub, Apigee, Cloud Spanner/AlloyDB, Memorystore) or comparable enterprise cloud environments.
• Strong background architecting systems in regulated domains such as FinTech, Payments, Banking, Insurance or HealthTech, where compliance, explainability and data isolation are essential.

Other open roles.

Treasury Senior Associate – Asset Management — Mumbai · Hybrid

Teradata Developer — Pan-India · Hybrid

MPM Associate — Mumbai · Hybrid

See all open roles →