AI Optimization Engineer

Company introduction We are building the optimization engine at the core of a new Capital Investment Planning & Prioritization Tool (CIPT). CIPT helps transmission, distribution, and vegetation planning teams decide which capital investments to fund, when to fund them, and why. Role summary / purpose AI Optimization Engineers design, build, test, and support optimization models and services that turn investment value, cost, risk, and constraint data into defensible, reproducible capital plans. In this role, you will formulate and implement Mixed-Integer Linear Programming (MILP) models for portfolio optimization, connect them to production cloud services on AWS, and work with data and ML engineers whose asset-health predictions feed the optimizer. You will also learn how GitHub Copilot, AI agents, and Spec-Driven Development (SDD) can improve the way optimization software is designed, developed, tested, and delivered. Strong engineering and modeling fundamentals remain important—planners, finance partners, and regulators rely on results, so each solution must be mathematically correct, deterministic, explainable, and auditable. Key responsibilities Optimization Modeling & Solution Development • Translate capital planning business rules into mathematical optimization models, including budget limits, resource and cash-flow limits, jurisdictional and portfolio constraints, must-do vs. discretionary planning groups, parent-child investment bundles, investment alternatives, and timing windows. • Formulate, implement, and tune MILP models using Python modeling frameworks such as Pyomo or PuLP, with commercial and open-source solvers such as Gurobi, HiGHS, or CBC. • Implement weighted value functions that combine benefit measures (reliability, safety, environmental, financial, PV/NPV, lifecycle value curves) into optimization objectives. • Design and build the scenario optimization workflow: create, copy, and optimize scenarios across the planning cycle (Initial, Pre-Opt, Post-Opt, Budget Lock), plus iterative and ad hoc reruns and manual overrides. • Build a deterministic run harness that guarantees reproducible results through solver settings, seeds, tolerances, and versioned input snapshots. • Develop infeasibility and diagnostic capabilities so planners understand why a scenario failed or why an investment was or was not selected. • Evaluate solvers and modeling approaches by comparing solution quality, run-time performance, licensing cost, and operational fit, and recommend options. • Integrate predictive model outputs (Probability of Failure, Remaining Useful Life, reliability consequence, risk exposure) from the asset predictive analytics module as inputs to the optimization. • Package and run optimization workloads as scalable cloud services (AWS Batch on Fargate, Step Functions, Lambda, containerized Python services) with REST APIs. • Use Infrastructure as Code approaches such as Terraform to build and maintain optimization infrastructure. • Use GitHub and GitHub Copilot as part of the software development lifecycle. • Gain hands-on experience with AI agents, agent integrations, orchestration patterns, and automation workflows. • Contribute reusable optimization components, model templates, test harnesses, and engineering accelerators that other teams can use. Optimization Validation, Performance & Explainability • Build automated test suites for optimization models, including unit tests for constraints, golden-datas…