Technical Lead – Generative AI

Job Title & Location

Generative AI Technical Lead

Dallas, TX – Onsite 4 days a week

The company is headquartered in Dallas, TX, and is a technology and strategy consultancy that aims to provide a competitive edge for its clients by solving complex problems with data, software, and strategy. It specializes in technology strategy, product development, software engineering, and digital transformation, with a particular emphasis on AI, MLOps, and Data Engineering. The firm’s clientele spans various industries, including AgTech, Healthcare, Logistics, and Financial Services.

You will work with technologies that include RAG, Agentic AI, Python, and LLMOps.

$120,000 – $145,000

• Own the end-to-end technical quality of generative AI systems — from data preparation and retrieval infrastructure through model integration, prompt design, output evaluation, deployment, and production monitoring.
• Establish and enforce GenAI engineering standards across the team: prompt versioning and management, evaluation harness design, context window strategy, output quality testing, hallucination tracking, and system documentation requirements.
• Make final technical decisions on LLM selection, context architecture, retrieval strategy, fine-tuning approaches, and orchestration frameworks — with the judgment to know when a simpler system outperforms a complex one.
• Own the technical risk register for each engagement — identifying context poisoning risks, hallucination failure modes, latency bottlenecks, cost overruns, and compliance exposure before they surface in production.
• Design end-to-end GenAI system architectures that integrate LLMs cleanly with enterprise data platforms, application layers, and operational workflows — built for reliability, observability, and controlled evolution.
• Architect retrieval-augmented generation (RAG) systems with rigorous attention to chunking strategy, embedding model selection, vector store design, retrieval quality evaluation, and reranking — treating retrieval as an engineering discipline, not an afterthought.
• Design agentic AI systems with well-defined tool interfaces, error handling, state management, and human-in-the-loop controls — architectures that behave predictably under real enterprise data and user behavior.
• Architect LLMOps foundations covering model gateway management, prompt registry, evaluation pipelines, A/B testing for prompts and models, cost monitoring, and production observability with output quality tracking.
• Lead and mentor a team of AI engineers and data scientists — setting technical direction, unblocking delivery, and raising the engineering quality of every individual contributor on the engagement.
• Represent the technical voice of the GenAI team in client-facing settings — communicating system behavior, failure modes, cost implications, and production risks with precision and candor.
• Establish incident response procedures for GenAI systems — owning the technical response when output quality degrades, retrieval pipelines drift, context windows overflow, or serving infrastructure fails under load.
• Ensure all GenAI systems meet client data governance, privacy, and compliance requirements — including data residency, PII handling in context, audit logging, and prompt injection defense at the architecture level.

• 7+ years in software or ML engineering; 3+ years with direct hands‑on ownership of production generative AI or LLM systems at enterprise scale.
• Deep production experience with LLM integration patterns — RAG architectures, function calling, tool use, structured output generation, and multi‑turn conversation management — beyond API wrappers and demo‑grade implementations.
• Strong engineering foundation in Python, software design principles, testing practices, and the discipline to build GenAI systems that engineering teams can operate, debug, and maintain without the original author present.
• Proven hands‑on experience with orchestration frameworks such as Lang Chain, Llama Index, or Lang Graph, and vector databases including Pinecone, Weaviate, pgvector, or Chroma in production retrieval systems.
• Demonstrated…