Hitachi

Agentic Systems Architect

In short

This role designs and deploys end-to-end agentic systems, bridging LLM/agent infrastructure (MCP, multi-agent orchestration, tool calling) with data architecture (Databricks lakehouses, semantic layers, knowledge graphs) at Hitachi Digital Services. It covers architecting MCP servers, vector databases, evaluation frameworks, observability/cost tracking, and secure deployment for agentic workflows. It requires 10+ years in software/platform/systems architecture and 2+ years shipping production LLM or agent applications, with hands-on expertise in multi-agent frameworks, knowledge graphs or vector databases, Python, Terraform, CI/CD and AWS. The position is hybrid, based in Dallas, Texas.

Function
Cloud & Data Engineering
Our Company
We're Hitachi Digital Services, a global digital solutions and transformation business with a bold vision of our world's potential. We're people-centric and here to power good. Every day, we future-proof urban spaces, conserve natural resources, protect rainforests, and save lives. This is a world where innovation, technology, and deep expertise come together to take our company and customers from what's now to what's next. We make it happen through the power of acceleration.
Imagine the sheer breadth of talent it takes to bring a better tomorrow closer to today. We don't expect you to 'fit' every requirement – your life experience, character, perspective, and passion for achieving great things in the world are equally as important to us.
Job Description
Summary
Design and deploy end-to-end agentic systems that integrate data, analytics, and operations. Bridge agent infrastructure (LLMs, tools, memory, MCP) with data architecture (lakehouses, semantic layers, knowledge graphs) to enable safe, governed, cost-controlled reasoning over complex data.
Key Responsibilities
•Design multi-agent orchestration (orchestrator-worker, hierarchical) and justify architecture for business use cases.
•Build constraint-driven agents with explicit guardrails, scoring gates, and approval workflows for high-impact actions.
•Architect MCP servers for enterprise integration; define auth, role-based access, stateful vs stateless patterns.
•Design semantic layers over Databricks (metrics, dimensions, business logic); implement text-to-SQL with cost estimation and query safeguards.
•Design knowledge graphs or ontologies for supplier/parts relationships; plan entity resolution and multi-hop queries.
•Select vector database (OpenSearch, Pinecone, native Databricks); design chunking, hybrid retrieval, metadata filtering.
•Establish tracing for agents: inputs, reasoning, tool calls, latency, tokens, cost, model versions.
•Design evaluation frameworks as release gates (golden sets, LLM-as-judge, regression tests).
•Implement per-decision cost tracking, alerts for runaway loops, per-tenant budgets.
•Design secure deployment (identity, secrets, compute), versioning, canary rollouts, on-call runbooks.
Required Experience
•10+ years in software engineering, platform engineering, or systems architecture.
•2+ years shipping production LLM or agent applications (not prototypes).
•Hands-on with multi-agent frameworks; understand loops, tool calling, memory, constraints.
•Built and deployed MCP servers; understand transport, auth, statefulness.
•Shipped evaluation frameworks (golden sets, LLM-as-judge) as release gates; measured agent quality in production.
•Designed or worked with lakehouse (AWS, Databricks or similar); understand semantic layers, text-to-SQL, governance.
•Production experience with vector databases; know ANN, HNSW, chunking trade-offs or Built or maintained knowledge graphs (RDF/OWL, Neo4j, property graphs); understand entity resolution and relationship inference.
•Built observability systems for multi-step workflows: traced tokens, latency, cost, model versions (OpenTelemetry, Dynatrace, etc.).
•FinOps or cost optimization in cloud systems; demonstrated ability to control runaway compute or API costs.
•Python, REST API design, infrastructure-as-code (Terraform), CI/CD, cloud infrastructure (AWS).
Nice to Have
•Databricks experience (Genie, Mosaic AI Agent Framework, Unity Catalog).
•Automotive or supply-chain domain expertise.
•GraphRAG or multi-hop reasoning patterns.
•Model routing and cost optimization (Haiku vs Opus).
•Human-in-the-loop approval workflows.
•Prompt versioning and systematic evaluation.
About Us
We're a global team of innovators. Together, we harness engineering excellence and passion to co-create meaningful solutions to complex challenges. We turn organizations into data-driven leaders that can make a positive impact on their industries and society.
Fostering innovation through diverse perspectives
Hitachi is a global company operating across a wide range of industries and regions. One of the things that sets Hitachi apart is the diversity of our business and people, which drives our innovation and growth. We are committed to building an inclusive culture based on mutual respect and merit-based systems.
How we look after you
We help take care of your today and tomorrow with industry-leading benefits, support, and services that look after your holistic health and wellbeing. We're also champions of life balance and offer flexible arrangements that work for you (role and location dependent).

Stop applying.
Get referred.

$10 in free credits

In just 60 seconds, fully on autopilot — Eraldt's AI agents read your CV, map the roles you fit, find live vacancies and the people who can refer you, write personal emails, and get you referred.

This isn't an application: your CV doesn't go to the company, and we don't promise a referral for this exact role.

Discover who else lands interviews on autopilot