AWS AI Architect - Agentic AI, Knowledge Graphs

Job Description

AWS AI Architect – Agentic AI, Knowledge Graphs

\n

Contract | United States

\n

 

\n

About EazyML

\n

EazyML ( is a Gartner-recognized AI company specializing in Responsible AI, explainability, and enterprise AI automation. We help organizations deploy AI solutions that are transparent, trustworthy, and designed to deliver measurable business value. Our team works across innovative AI technologies and enterprise-scale solutions, giving our professionals the opportunity to work on challenging, real-world AI initiatives.

\n

 

\n

The Opportunity

\n

EazyML is seeking an experienced AWS AI Architect -Agentic AI, Knowledge Graphs & Text-to-SQL to lead the development of a governed, enterprise-grade Agentic AI solution for a global enterprise. The solution will enable users to ask business questions in natural language and receive accurate, explainable, and validated answers across multiple data platforms.

\n

You will build on an existing proof of concept and evolve it into a secure, scalable, production-ready solution. This is a hands-on architecture role requiring strong engineering skills, deep SQL expertise, and practical experience with AWS-native AI services, agent orchestration, semantic layers, and knowledge graphs.

\n

You will own key architectural decisions, write critical production code, establish evaluation and quality standards, and lead an offshore team of data and GenAI engineers in collaboration with business and domain subject-matter experts.

\n

Key Responsibilities

\n

\n
• Agentic AI Architecture: Extend the existing Amazon Bedrock AgentCore and Strands Agents architecture to deliver a reliable, scalable, and production-ready AI solution.
\n
• Semantic Layer Development: Design and implement a governed business facts library covering join conditions, metric definitions, data grain, aggregations, and reusable business entities. Partner with subject-matter experts to validate and approve business definitions.
\n
• Ontology and Knowledge Graphs: Design and implement an ontology and knowledge graph using Amazon Neptune, enabling identity-based retrieval, relationship discovery, and enrichment with business metadata.
\n
• Deterministic Text-to-SQL: Build a reliable intent-to-SQL compiler that translates natural-language questions into validated SQL
\n
• Agent Orchestration: Integrate the semantic layer and SQL compiler into Strands Agents workflows. Develop robust probing, validation, and control mechanisms that go beyond prompt engineering to improve reliability.
\n
• Evaluation and Quality Assurance: Build an evaluation framework with golden datasets,
\n
• AWS Backend Engineering: Configure a production-oriented backend in a non-production environment, including DynamoDB for chat sessions and Amazon RDS for agent context and metadata.
\n
• Technical Leadership: Lead and mentor offshore data and GenAI engineers, coordinate implementation activities, and collaborate with business stakeholders and subject-matter experts to ensure technical solutions meet real-world requirements.
\n

\n

Required Qualifications

\n

\n
• 8+ years of experience in software engineering, data engineering, machine learning, or related technical disciplines, including at least 3 years designing and implementing AWS-based solutions.
\n
• Hands-on production experience with Amazon Bedrock, including foundation models, Knowledge Bases, and Guardrails. Experience with Bedrock AgentCore is a strong advantage.
\n
• Demonstrated experience building agentic AI workflows using Strands Agents, LangGraph, CrewAI, or comparable orchestration frameworks.
\n
• Practical experience designing ontologies and engineering knowledge graphs, preferably with Amazon Neptune and technologies such as RDF/SPARQL, Gremlin, or openCypher.
\n
• Strong SQL expertise and experience developing Text-to-SQL systems, semantic layers, or equivalent data-querying solutions, including join-path design, data grain, aggregation correctness, and business metric definitions.
\n
• Hands-on experience working across Snowflake, Databricks, and PostgreSQL.
\n
• Strong Python development skills and experience building evaluation frameworks, golden datasets, test harnesses, and accuracy metrics for LLM-based applications.
\n
• Working knowledge of AWS services and architecture, including DynamoDB, Amazon RDS, IAM, Lambda, and CloudWatch.
\n
• Understanding of secure software development, access controls, AI guardrails, and enterprise deployment practices.
\n
• Proven ability to make architectural decisions, write production-quality code, and communicate technical trade-offs clearly to stakeholders.
\n
• Experience leading or coordinating distributed engineering teams.
\n

\n

Preferred Qualifications

\n

\n
• Familiarity with semantic-layer technologies such as dbt Semantic Layer, Cube, or AtScale.
\n
• Experience with metadata catalogues such as AWS Glue Data Catalog or Amazon DataZone.
\n
• Knowledge of LLM observability, tracing, evaluation, and monitoring tools
\n
• Relevant AWS certifications, such as AWS Certified Solutions Architect – Professional or certifications in AI and machine learning.
\n

\n

Why Join EazyML?

\n

\n
• Work on a challenging, enterprise-scale Agentic AI initiative with a strong emphasis on accuracy, governance, security, and measurable business outcomes.
\n
• Take ownership of core architecture and hands-on implementation, with the opportunity to shape a solution from proof of concept to production readiness.
\n
• Collaborate with experienced engineers, and business stakeholders on advanced AI engineering challenges.
\n
• Opportunity for follow-on contract work as the solution expands across additional business domains, data sources, and a growing user base.
\n

\n

Engagement: US-based contract opportunity. Candidates should have strong hands-on engineering capabilities and be comfortable taking ownership of both architecture and implementation.

\n