• Translate user needs and strategic challenges into tractable ML problems. Establish baselines, define success criteria, and prioritize experiments by potential impact, data readiness, and feasibility.
• Develop models for classification, entity resolution, anomaly detection, forecasting, ranking, and risk assessment using text, geospatial, energy, market, media, and organizational data. Combine sources to reveal patterns that improve decisions.
• Select and test methods suited to the problem, from statistical models and classical ML to deep learning, graph methods, and LLMs. Demonstrate when added complexity delivers meaningful value.
• Use predictive models, optimization, and scenario analysis to evaluate options and allocate resources. Apply experimental design and causal inference where the data and assumptions support credible conclusions about intervention effectiveness.

Data Quality and Scientific Evaluation

• Build representative datasets, labeling strategies, and evaluation benchmarks. Prevent leakage and test performance across time, geography, groups, and unfamiliar contexts using validation appropriate to the intended use.
• Quantify uncertainty, calibration, and the consequences of false positives and false negatives. Develop approaches for limited or noisy labels, and define when systems should defer to human judgment.
• Make research reproducible through versioned data, code, and experiments. Document assumptions, limitations, and failure modes; monitor drift and use feedback to guide model improvement.

LLM and Agentic Systems

• Build retrieval and synthesis systems that turn heterogeneous documents and data into verifiable answers and actionable intelligence. Improve retrieval, ranking, source quality, citation accuracy, and faithfulness through systematic evaluation.
• Develop agents that use tools, models, and internal services to complete useful workflows. Define permissions, failure recovery, and human review for consequential actions, and evaluate complete task outcomes.
• Evaluate hosted and open-weight models, prompting, fine-tuning, and distillation against quality, privacy, latency, and cost requirements. Build reusable capabilities where they offer an advantage across use cases.
• Establish evaluation and monitoring for LLM and agent systems, including factual accuracy, robustness, task completion, and operating cost. Use controlled experiments to verify improvements before release.

Production Integration and Shared Capabilities

• Shape reusable data representations, feature pipelines, retrieval services, and model interfaces. Introduce ontologies or knowledge graphs where relationships and structured reasoning improve performance or usability.
• Partner with software and ML engineers to move validated prototypes into reliable products. Define data and API contracts, deployment requirements, monitoring, and retraining plans while balancing performance, maintainability, and cost.

Technical Direction and Measurable Impact

• Identify high-value opportunities beyond existing platform roadmaps. Propose new capabilities and research directions, define milestones and decision points, and help prioritize the AI/ML portfolio.
• Work with product, research, communications, and program teams to understand user decisions and constraints. Explain results clearly, challenge unsupported assumptions, and help colleagues apply AI/ML effectively.
• Define and track outcomes such as decision quality, predictive performance, user adoption, analyst time saved, and cost per workflow. Use evidence to scale successful approaches, revise weak ones, and stop work that does not deliver value.

Required Qualifications

• PhD in machine learning, computer science, statistics, or a related quantitative field, with 3-5 years of applied ML experience. Exceptional recent PhD graduates with substantial applied work will also be considered.
• Evidence of turning ambiguous problems into rigorous experiments and useful AI/ ML capabilities, including appropriate baselines, validation, uncertainty assessment, and error analysis.
• Hands-on experience with NLP and LLM applications, including embeddings, retrieval-augmented generation, and systematic evaluation.
• Strong Python skills and experience with tools such as PyTorch/ TensorFlow, scikit-learn, and pandas; proficiency in SQL, version control, and reproducible research. Experience deploying and evaluating models in real workflows, independently or with engineering partners.
• Sound statistical judgment and the ability to work with heterogeneous, incomplete, or noisy data. Ability to distinguish predictive associations from evidence of causation.
• Ability to own technical work, prioritize under uncertainty, and communicate findings to technical and nontechnical colleagues. Curiosity about energy and climate challenges and motivation to learn from domain experts.

Preferred Qualifications

• Experience with energy, climate, grid, infrastructure siting, permitting, regulatory, or public participation data; geospatial analysis, graph ML, or time-series forecasting.
• Publications, open-source contributions, or deployed projects demonstrating methodological depth, originality, and practical impact.

Compensation, Benefits, and Perks

• Annual salary range: $160,000-190,000, commensurate with experience
• Medical, dental, and vision insurance with employee and dependent premiums paid at 95-100% by GoodPower.
• 2-4 weeks of paid vacation annually, depending on seniority
• 13 paid holidays and 2 floating holidays
• 2 weeks of US office closure
• 10 days of paid sick leave annually.
• 403(b) with 3% employer contribution/match.
• Charitable giving: 1:1 match for charitable contributions ($250 per year), plus 2 paid days off for volunteering.
• Education reimbursement up to $1,500 per year.
• Healthcare Flexible Spending Account (FSA) and Health Savings options
• Stipend of $200/month (for cell phone, wellness expenses, etc.)
• 8 weeks of paid sabbatical after 5 years of employment
• 8 weeks of paid parental leave.