Apply on Employer Site

Chevron · 2 days ago

Principal Applied Scientist - Enterprise AI

Houston, TX

Full-time

Onsite

Lead/Staff

10+ years exp

Chevron is seeking a Principal Applied Scientist to lead the design and delivery of industry-specific foundation models and agentic AI systems. The role involves architecting AI foundations, mentoring teams, and developing reusable frameworks to support Chevron's AI ambitions at a global scale.

EnergyFossil FuelsGeothermal EnergyManufacturingNatural ResourcesOil and GasRenewable Energy

No H1B

Responsibilities

Architect Industry Foundation Models Design domain tailored FMs for energy workflows (e.g., subsurface, production operations, supply & trading), leveraging new and emerging FM architectures and advanced methods such as instruction tuning, LoRA, adapter based finetuning, prompt optimization, and RLHF/RLAIF

Fine Tune & Optimize Model Weights Own the end to end process for weight initialization, adaptation, quantization, distillation, and evaluation to meet accuracy, latency, and cost targets; establish best practices for versioning, reproducibility, and model cards

Develop & Operationalize AI Systems Build, test, and deploy production grade models and agentic workflows using AzureML, Azure OpenAI Service, and Databricks (Unity Catalog & Model Serving); maintain flexibility to integrate approved third party platform and/or model endpoints via Azure where appropriate, e.g. DataRobot

Establish Reusable AI Frameworks Create modular, reusable components—feature stores, orchestration pipelines, evaluation harnesses, prompt/agent templates, and retrieval layers—that accelerate delivery across multiple AI applications

Technical Leadership & Mentorship Provide deep technical guidance to data scientists, AI engineers, and software engineers; lead design reviews, code reviews, and drive adoption of best practices for scalable AI

Collaborate Across Disciplines Partner with business stakeholders and cross functional engineering delivery teams to translate complex problems into measurable, scalable AI solutions, and to embed models into production systems and agentic workflows

Model Lifecycle & Responsible AI Work alongside Chevron’s Data & Insights Department to define standards for governance, monitoring, drift detection, retraining, and evaluation (offline/online); champion fairness, transparency, safety, and auditability across the model lifecycle

Innovation & Thought Leadership Lead research and experiments on foundation models, multimodal learning, retrieval augmented generation, and reinforcement learning; publish findings internally and externally where appropriate

Qualification

Foundation modelsPythonAzureMLMachine learning frameworksEnergy workflowsMLOpsGenerative AIReinforcement learningOptimization methodsDistributed model trainingTechnical leadershipCross-functional collaborationCommunication skills

Required

Advanced degree (MS or PhD) in computer science, statistics, mathematics, neurosciences, machine learning, engineering or related quantitative field

10+ years building and deploying enterprise-scale AI/ML systems, including technical leadership of complex initiatives

Demonstrated expertise in foundation models and open-source LLM/SLM ecosystems, including fine-tuning, weight management, and evaluation

Strong proficiency in Python and modern ML frameworks (PyTorch, TensorFlow, Hugging Face)

Hands-on experience with AzureML, Azure OpenAI Service, and Databricks

Deep domain understanding of energy workflows (upstream, downstream, or supply & trading) and the ability to encode domain insights into model design

Excellent communication, cross-functional collaboration, and the ability to set technical strategy and mentor teams

Preferred

Experience with multimodal FMs (text, time-series, tabular, vision) and RAG systems

Experience with agentic AI architectures, generative AI, and reinforcement learning

Familiarity with MLOps, CI/CD for ML, model serving, experiment tracking, and online evaluation

Demonstrated background in optimization methods, causal inference, or simulation/agent-based modelling for decision support

Demonstrated experience in distributed model training on multi-GPU and multi-node clusters and ability to optimize and troubleshoot training across a cluster to ensure efficient utilization of resources and consistent model performance

Record of technical publications, patents, or conference presentations in AI/ML