Gen AI Architect

Job Title: GenAI Architect (Agentic AI)

Houston, TX hybrid

Role Context

The Advisor Relationship

This role serves as the primary technical authority supporting the Head of AI Products. While the client defines product strategy and commercial direction, the AI Architect owns the technical depth—translating vision into architecture, validating proposals with evidence, and ensuring all AI systems are built on strong foundations.

• Depth Expected:
• Hands-on capability to build PoCs, review production architectures, and guide engineering teams.
• Responsibilities:
• Architecture decisions
• Technical client advisory
• Prompt engineering standards
• Evaluation quality
• Model selection
• Security review
• Escalation to Client:
• Roadmap priority conflicts
• Budget decisions
• Executive relationships
• Organizational policies
• Client Exposure:
• Senior engineers, CTOs, and occasional C-suite stakeholders

What You Own

Technical Leadership

• Design agentic AI architectures (orchestration, tool-calling, memory, state management)
• Evaluate and select LLMs (cost, latency, capability, data residency)
• Own end-to-end RAG pipeline quality (chunking, embeddings, retrieval, re-ranking, generation)
• Define prompt engineering standards, versioning, and regression practices
• Establish evaluation frameworks and production quality gates
• Define security posture (prompt injection defense, PII handling, audit logging, VPC)
• Review MCP tool schemas, authentication, and authorization patterns

Advisory & Enablement

• Act as technical advisor in client workshops and architecture reviews
• Translate technical trade-offs into business impact
• Guide AI adoption (intent classification, agent design, vector search)
• Build reference architectures and decision frameworks
• Coach engineers on AI best practices (evals, observability, CI/CD)
• Stay updated on AI trends, frameworks, and risks
• Build PoCs and demos to validate technical direction

Core Skill Matrix

AI Architecture

• Agentic Design: Orchestration, sub-agents, memory, failure handling (Expert | Owns)
• LLM Selection: Cost, latency, privacy trade-offs (Strong | Owns)
• RAG & Retrieval: End-to-end pipeline design (Expert | Owns)
• MCP & Tool Calling: Schema design and safety validation (Strong | Owns)
• ADK Frameworks: LangGraph, CrewAI, AutoGen, Google ADK (Strong | Owns)
• Vector Search: Pinecone, pgvector, Qdrant (Strong | Owns)
• Intent Classification: Routing, semantic classification (Solid | Owns)

Prompting & Evaluation

• Prompt engineering (CoT, few-shot, structured outputs) (Expert | Owns)
• Evaluation frameworks & LLM-as-judge (Expert | Owns)
• Fine-tuning strategy (SFT, LoRA) (Solid | Leads)

Security & Infrastructure

• Security posture (PII, injection defense, guardrails) (Strong | Owns)
• Networking & infra (VPC, IAM, endpoints) (Solid | Owns)
• Cost optimization (token usage, batching, caching) (Strong | Leads)

Advisory

• Technical-to-business translation (Expert | Owns)
• Client workshops & discovery sessions (Strong | Leads)
• Reference architectures & playbooks (Strong | Owns)

Effectiveness

• Ambiguity handling (Strong | Owns)
• Continuous learning (Expert | Owns)
• Transparent communication (Expert | Owns)

What We Look For

Must-Have Qualifications

• 5 years in AI/ML engineering, solution architecture, or AI consulting
• Proven experience delivering production-grade AI/LLM systems
• Hands-on experience with ≥2 AI frameworks and vector databases
• Strong Python skills (PoCs architecture reviews)
• Experience designing/debugging enterprise RAG pipelines
• Ability to engage with senior engineers and leadership stakeholders