Lead AI Architect ( Agentic SDLC )

Orion Innovation•Edison, NJ

About The Position

Orion Innovation is seeking a hands-on applied AI engineering leader to design and build agentic workflows that support the software development lifecycle. The role connects requirements, architecture, implementation, testing, code review, deployment, and maintenance, turning prototypes into dependable systems that engineering teams adopt and trust. The position combines hands-on engineering, technical architecture, product judgment, and customer-facing leadership. The successful candidate will work with engineering leaders, developers, and customers to find high-value opportunities, deliver working solutions, and assess their impact. They will also guide teams from early experimentation through production rollout and use deployment evidence to shape product direction and reusable engineering practices.

Requirements

At least 5 years of experience in software engineering, applied AI, developer tooling, platform engineering, or technically intensive solutions engineering, with a strong record of shipping working systems.
Strong programming skills in Python, TypeScript, or another widely used language, and confidence navigating unfamiliar codebases.
Deep practical experience with AI coding tools, including concrete examples of workflows, integrations, or agents built and used.
Hands-on experience building LLM applications with tool calling, structured outputs, context retrieval, and iterative execution.
Strong understanding of software engineering fundamentals, including architecture, Git, automated testing, code review, CI/CD, debugging, and production operations.
Judgment to distinguish tasks suited to agents from those better handled by deterministic automation, and to determine where human review is essential.
Disciplined approach to evaluating generated code and diagnosing failures across models, prompts, context, tools, and execution environments.
Clear written and verbal communication skills, including the ability to explain architectural tradeoffs and lead hands-on technical sessions.
Comfort with ambiguity, fast experimentation, and ownership from problem discovery through delivery and adoption.

Nice To Haves

Built agents that complete multi-step engineering tasks against real repositories.
Developed agent evaluation frameworks, repository context systems, custom tools, MCP integrations, or orchestration infrastructure.
Integrated coding agents into enterprise environments with meaningful security and operational requirements.
Contributed developer tools, open-source projects, technical guides, or reference implementations.
Demonstrated measurable improvements in engineering throughput or quality without shifting hidden work onto reviewers.
Strong product instinct: identifies valuable problems, builds the smallest useful solution, and improves it based on evidence.

Responsibilities

Build coding agents that understand repositories, implement features, diagnose defects, write and run tests, address review feedback, and prepare changes for human approval.
Design coordinated agentic workflows across issue trackers, repositories, developer environments, CI/CD pipelines, documentation, and observability tools, with clear ownership and approval points.
Engineer reliable agent execution, including context management, tool integrations, orchestration, isolated execution environments, state persistence, retries, and recovery mechanisms.
Use AI coding tools such as Codex, Cursor, and Claude Code to prototype, implement, debug, and refine systems, and turn effective practices into repeatable team capabilities.
Create prototypes, reusable integrations, reference architectures, and automations that demonstrate value and support adoption.
Develop benchmarks and evaluation harnesses to measure task completion, code correctness, regression rates, security findings, latency, cost, and engineering effort.
Implement scoped permissions, secrets handling, audit trails, sandboxing, and human review controls to make agent autonomy safe and practical.
Work with engineering teams to understand constraints, run hands-on workshops, and guide adoption from initial experiments through production rollout.
Translate deployment experience into actionable product requirements, model feedback, technical guides, and reusable engineering patterns.
Demonstrate measurable outcomes and share a repository, live demo, technical walkthrough, or concrete example of an agentic development workflow.