AI Engineer Team/Project Lead
Job Description
Halliburton is hiring an AI Engineering Team/Project Lead in Texas (onsite) to lead hands-on development of production AI agents for engineering and operations workflows on an established Databricks and Azure agent platform.
Responsibilities
- Partner with a project team of two other engineers, write production code daily, and focus on the hardest technical problems.
- Lead the team in turning real engineering and operations workflows into production agents, from initial discovery to daily use.
- Collaborate with business stakeholders and the engineers doing the work to define requirements, cut scope, and maintain a roadmap centered on highest-value delivery.
- Define agent architecture, review designs and code, and enforce a high bar for reliability, security, and maintainability.
- Integrate secure tool and data connectors (including MCP servers) to provide agents governed access to Halliburton data and engineering applications.
- Use platform evaluation and tracing to demonstrate agent reliability, then track adoption and time saved in production.
- Build optimization routines and monitoring to detect drift, regressions, and failures early, keeping agents aligned with intended behavior after launch.
- Establish AI-first development practices using coding agents, shared templates, and code reviews that help every engineer ship faster.
- Own ongoing improvements: fix what breaks and iterate agents based on real usage.
- When gaps appear, contribute reusable components such as templates, connectors, or evaluation so the next workflow delivers faster.
- Lead from the keyboard on workflow delivery and implementation, staying hands-on with production engineering.
Requirements
- 6+ years building and shipping production software (or equivalent demonstrated work).
- Degree in Computer Science, Data Science, or a related technical field (or equivalent experience).
- Proven experience leading technical work for a small engineering team (architecture, code review, mentoring) while remaining a hands-on contributor.
- Track record working with business stakeholders to convert ambiguous needs into clear requirements and a delivery roadmap.
- Strong Python skills and comfort across the stack, including APIs, services, and simple internal user interfaces.
- Shipped LLM-powered agents to real users, ideally multi-agent systems (not only single prompt chains), including tool calling, RAG, structured outputs, and prompt/context engineering.
- Hands-on experience with agent frameworks and protocols such as: LangGraph, Microsoft Agent Framework / Semantic Kernel, OpenAI Agents SDK, and Model Context Protocol (and judgment to know when not to use one).
- Experience testing non-deterministic systems, including building eval sets, automated scoring, and regression checks.
- Solid engineering fundamentals: Git, automated testing, CI/CD, containers, and cloud deployment.
- Hands-on Azure experience, including experience with one or more: Microsoft Foundry (Azure AI Foundry) Agent Service, Azure OpenAI, Azure AI Search, Azure Kubernetes Service, Container Apps, and Entra ID.
Technologies
- Python
- Databricks
- Azure
- MCP servers / Model Context Protocol (MCP)
- LangGraph
- Microsoft Agent Framework
- Semantic Kernel
- OpenAI Agents SDK
- Tool calling
- RAG
- Structured outputs
- Git
- CI/CD
- Containers and cloud deployment
- Microsoft Foundry (Azure AI Foundry) Agent Service
- Azure OpenAI
- Azure AI Search
- Azure Kubernetes Service
- Container Apps
- Entra ID
- Coding agents
Nice to have
- Databricks: Mosaic AI Agent Framework / Agent Bricks, MLflow tracing and evaluation, Unity Catalog, Model Serving, Vector Search.
- Developer tooling: contributions to SDKs, internal platforms, or shared libraries used by other engineers.
- Total ownership: taking work from idea to production and staying accountable for outcomes.
- Open source: contributions to AI or developer tooling projects.
- Energy or industrial exposure (helpful, not required).
What success looks like
- First 30 days: understand the platform and your team; align on the team’s first workflow, its success measure, and an initial roadmap with business stakeholders.
- First 90 days: ship the team’s first workflow and have it in regular use by the intended users, with a measured result such as hours saved; contribute a meaningful share of the code.
- First 6 months: operate several workflows in daily use; stakeholders trust the roadmap; architecture and review standards enable faster shipping.
Location
- Houston, Onsite
- Texas (onsite)