Reinforcement Learning & Controls Engineer

EQORE Inc.; Somerville, Northern (MA, KY); Hybrid; USD 120,000 - 180,000; Full time

Job summary

EQORE Inc. seeks a hands-on ML engineer to own QoreAI's learning-based control and optimization stack, designing controllers for energy systems and evaluating performance under real-world constraints.

You will build reinforcement learning pipelines in Python, integrate forecasts and market signals, and ensure safe, scalable deployment as the fleet grows across sites and tariffs. You’ll work in person with a small team and contribute to reliable, measurable improvements that customers see on

Qualifications

• Hands-on experience building reinforcement learning systems for continuous control or sequential decision-making.
• Strong Python and experience owning complete reinforcement learning training and evaluation pipelines.
• Strong command of deep RL fundamentals, including policy optimization, exploration, credit assignment, stability, and sample efficiency.
• Experience designing environments and rewards around noisy real-world data, partial observability, and operational constraints.
• Understanding of safe deployment, failure modes, and rigorous evaluation in closed-loop systems.

Responsibilities

• Own QoreAI's learning-based control and optimization architecture.
• Design and evaluate control approaches for battery operation under uncertainty and real-world constraints.
• Develop, train, tune, and evaluate learning-based control policies against strong heuristic and optimization baselines.
• Build simulation, offline evaluation, backtesting, shadow-mode, and production-validation systems.
• Integrate forecasts and market signals produced with the Energy Markets engineer into control inputs and evaluation environments.
• Support ConnectedSolutions, Clean Peak Standard, ICAP, and other programs.
• Generalize controllers across sites, tariffs, grid programs, operating regimes, and hardware.
• Monitor live performance, diagnose regressions and distribution shift, and build safe fallbacks.

Skills

Reinforcement learning

Python programming

RL training pipelines

Closed-loop control

Policy optimization

Tools

PyTorch

JAX

Job description

Own learning-based control work from problem formulation through evaluation and deployment, turning new approaches into safe, measurable production performance.

About the role

QoreAI decides when each system charges and discharges across demand charges, tariffs, and grid programs. Our control stack combines optimization, simulation, and learning-based methods to improve performance across a growing fleet. You will turn new control approaches into safe, measurable production performance.

You will also maintain strong optimization baselines and safe fallbacks so we can measure improvements and operate reliably.

What you'll do

• Own QoreAI's learning-based control and optimization architecture.
• Design and evaluate control approaches for battery operation under uncertainty and real-world constraints.
• Develop, train, tune, and evaluate learning-based control policies against strong heuristic and optimization baselines.
• Build simulation, offline evaluation, backtesting, shadow-mode, and production-validation systems.
• Integrate forecasts and market signals produced with the Energy Markets engineer into control inputs and evaluation environments.
• Support ConnectedSolutions, Clean Peak Standard, ICAP, and other programs.
• Generalize controllers across sites, tariffs, grid programs, operating regimes, and hardware.
• Monitor live performance, diagnose regressions and distribution shift, and build safe fallbacks.

What you'll bring

• Hands-on experience building reinforcement learning systems, ideally for continuous control, sequential decision-making, or closed-loop operation.
• Strong Python and experience owning complete reinforcement learning training and evaluation pipelines.
• Strong command of deep RL fundamentals, including policy optimization, exploration, credit assignment, stability, and sample efficiency.
• Experience designing environments and rewards around noisy real-world data, partial observability, and operational constraints.
• Understanding of safe deployment, failure modes, and rigorous evaluation in closed-loop systems.
• A desire to work full-time and in person with a small team.

Nice to have

JAX or PyTorch, offline RL, model-based RL, imitation learning, energy, robotics, industrial control, mathematical optimization, or forecasting.

Why EQORE

Your controller will move power through real systems and its performance will appear on customer bills. As the fleet grows, every improvement compounds.