AI Application Architect – Enterprise AI Solutions

About the role

The AI Application Architect will lead the design of reference architectures, guardrails and delivery patterns for enterprise‑grade AI applications. You will own end‑to‑end architecture across application layers, data, model lifecycle, integration, security and operations, ensuring performance, cost efficiency and compliance.

Key responsibilities

• Define, document and evolve target‑state architecture for AI‑enabled services, including micro‑services, event‑driven and API‑first designs.
• Establish canonical integration patterns for LLM invocation (synchronous, streaming, async, batch) and system‑of‑record interactions.
• Design and implement LLM solutions using Retrieval‑Augmented Generation, embeddings, vector search and prompt orchestration, selecting appropriate hosted or open‑source models.
• Partner with Data Science and ML Engineering to decide on model selection, fine‑tuning, distillation and deployment strategies.
• Define platform capabilities for prompt/version management, model registries, feature stores and standardize CI/CD pipelines for AI.
• Collaborate with security, compliance, SRE and product teams to align architecture with business outcomes and governance.
• Evaluate emerging models, frameworks and tooling, run POCs and maintain a living reference architecture library.
• Enforce architecture fitness functions, infrastructure‑as‑code, policy‑as‑code and produce high‑quality documentation and runbooks.

Required profile

• 10+ years of experience in software/application architecture or senior engineering roles delivering complex distributed systems.
• 2–4+ years of hands‑on experience designing and delivering AI/LLM solutions.
• Proven ability to lead technical discussions, produce architecture decision records and present to governance boards.

Required skills

• Microservices architecture
• Event‑driven design
• API‑first development
• Large Language Model (LLM) integration
• Retrieval‑Augmented Generation (RAG)
• Embeddings and vector search
• Prompt engineering and orchestration
• Model registries and feature stores
• CI/CD pipelines for AI
• Observability (tracing, metrics, logging)
• Infrastructure‑as‑code
• Policy‑as‑code