AI Principal Architect
Salary not listed • Pune, India • Full-time
Posted 1d ago
Job Location
Pune, India
Tech Stack
Remote Work Policy
On-site
Employment Type
Full-time
Categories
AI Agent Engineer
About the job
Mastercard's Business & Market Insights (B&MI) group is seeking a Principal Architect, Generative AI & ML Engineering, to lead the architecture vision and technical strategy for their agentic AI platform. This platform enables agents to make extensive tool calls into enterprise services and APIs, requiring secure, reliable, and scalable operation. The role involves setting standards for tool calling, multi-agent orchestration, self-correcting agents, memory, retrieval, and LLMOps, while making key decisions on models, platforms, and trade-offs. You will collaborate with engineering, product, security, and business leaders to shape the long-term direction of Gen AI within B&MI and mentor other architects and engineers, while also staying hands-on with prototypes to de-risk significant initiatives. The focus is on solving complex enterprise Gen AI challenges related to reliability, integration, security, and scale, leveraging experience in backend or distributed systems architecture applied to LLM systems.
Responsibilities
- Own the architecture and technical roadmap for the Operational Intelligence Gen AI platform, defining reusable reference architectures and standards.
- Architect the enterprise tool-calling layer, establishing standards for exposing internal services, data platforms, and external APIs as tools for agents.
- Set the architecture for multi-agent and skills-based agent systems, defining coordination, delegation, and context sharing mechanisms.
- Design self-correcting agent patterns, including tool critics, verifiers, reflection, and plan-and-execute loops.
- Architect agent memory and state for long-running, multi-step sessions, covering short- and long-term memory, session persistence, and memory scoping.
- Define retrieval architecture for grounded reasoning, including Self-RAG, CRAG, adaptive retrieval, and Graph-RAG, utilizing vector databases and knowledge graphs.
- Design the inference and orchestration layer for high throughput, encompassing async processing, streaming, caching, rate limiting, and model routing.
- Establish the LLMOps architecture, including MLflow, CI/CD for prompts and agents, and model lineage.
- Set platform-wide evaluation and observability standards for tracing, accuracy, success rates, cost, and latency.
- Build security and Responsible AI principles into the architecture, including guardrails, prompt-injection defenses, hallucination mitigation, and data privacy.
- Own key architecture decisions and trade-offs, such as build versus buy and model/vendor selection, documenting through architecture decision records.
- Assess emerging models, frameworks, and standards for production readiness.
- Lead design reviews and partner with cross-functional teams to align on architecture and improve engineering standards.
- Build prototypes and reference implementations to de-risk decisions and provide starting points for teams.
Requirements
- Bachelor's or Master's degree in Computer Science, AI/ML, or Engineering, or equivalent practical experience.
- Significant experience architecting large-scale distributed systems in production (microservices, event-driven systems, data platforms, cloud-native infrastructure).
- Deep expertise in API design and systems integration, including contracts, security models, and reliability patterns for services consumed by multiple teams.
- Hands-on experience designing and shipping LLM-powered systems in production, particularly agents using tool/function calling, memory, and retrieval.
- Strong Python skills for hands-on prototyping (async programming, FastAPI or similar).
- SQL proficiency.
- Deep experience with at least one major cloud platform (AWS preferred; Azure or GCP welcome), including its AI services.
- Solid grounding in security, privacy, and compliance for enterprise environments.