AI Architect – Generative AI & Agentic AI

This is a remote position.

Role Overview

The AI Architect will lead the architecture, design, and production rollout of enterprise-grade AI solutions, with a focus on Generative AI (GenAI), Retrieval-Augmented Generation (RAG), and agentic AI systems. The role translates business needs into secure, scalable, reliable, cost-effective architectures and establishes reusable architecture patterns, engineering standards, and governance practices. The architect partners with product, data, platform, security, engineering, and business leaders to take AI capabilities—from conversational assistants and document intelligence to workflow automation and decision support—from experimentation into production.

Key Responsibilities

· Define enterprise AI architecture strategy, reference architectures, technology standards, and a pragmatic roadmap aligned to business outcomes.

· Design end-to-end AI solutions spanning data ingestion, model access, retrieval, orchestration, APIs, user experience, deployment, monitoring, and governance.

· Architect production GenAI, RAG, and agentic workflows with appropriate controls for quality, latency, availability, scalability, and cost.

· Assess and select foundation models, embedding models, vector/search technologies, agent frameworks, cloud services, and deployment patterns; document trade-offs and decisions.

· Establish reusable components, architecture patterns, design reviews, engineering guidelines, and platform capabilities that accelerate delivery across teams.

· Lead technical discovery, prototypes, architecture spikes, and production readiness reviews; guide teams through implementation and operational handover.

· Partner with security, privacy, legal, risk, and compliance stakeholders to implement responsible AI, data protection, access controls, auditability, and risk-based safeguards.

· Define evaluation and observability practices for AI systems, including quality, groundedness, safety, drift, latency, usage, and cost metrics.

Mentor engineers and technical leads; communicate architecture decisions and risks to executive, product, and engineering stakeholders.

Requirements

Mandatory Skills & Experience

1. Enterprise AI & GenAI Architecture

· Demonstrated experience designing and deploying production-grade GenAI applications such as conversational assistants, summarization, document intelligence, knowledge assistants, and decision-support systems.

· Strong understanding of LLM behavior and constraints: tokenization, context windows, prompting, structured outputs, tool/function calling, model selection, and failure modes.

· Experience designing inference and serving patterns for latency, throughput, and cost, including streaming, batching where appropriate, caching, asynchronous processing, and response chunking.

· Ability to evaluate build-versus-buy, hosted-versus-self-hosted, and model/platform trade-offs.

1. Retrieval-Augmented Generation (RAG) & Knowledge Systems

· Hands-on architecture experience with ingestion, parsing, metadata enrichment, chunking, embeddings, indexing, retrieval, reranking, context assembly, and grounded generation.

· Knowledge of dense retrieval, BM25, hybrid search, metadata filtering, semantic ranking, query rewriting, multi-hop retrieval, and retrieval evaluation.

· Experience with vector databases/search platforms such as pgvector, Pinecone, Weaviate, Milvus, Qdrant, Chroma, Elasticsearch/OpenSearch, or cloud-native equivalents.

· Design for enterprise knowledge sources, document permissions, freshness/synchronization, provenance, feedback loops, and measurable retrieval quality.

1. Agentic AI & Workflow Architectures

· Experience architecting tool-using agents and multi-agent/workflow systems, with clear boundaries for planning, execution, delegation, state, memory, and human approval.

· Familiarity with frameworks such as LangGraph, LangChain, Microsoft Agent Framework / AutoGen, Semantic Kernel, and CrewAI; select frameworks based on use case rather than framework preference.

· Design reliable orchestration using state machines, durable workflows, event-driven patterns, retries, timeouts, idempotency, and human-in-the-loop controls.

· Knowledge of tool/function calling, MCP (Model Context Protocol), API integration, agent-to-agent communication patterns, and secure tool authorization.

· Understand planning-and-execution, reflection/critique patterns, memory approaches, and the limits of…