Applied Scientist – Solar Analytics

Swish Solar is building the operating layer for utility scale solar infrastructure. As solar farms scale globally, owners and operators need better systems to understand solar asset performance.

Our core platform, SwishOS, brings together SCADA, inverter, weather, satellite, soiling, and operational data to give solar teams a real-time view of plant performance and maintenance priorities. SwishOS helps operators identify energy losses, estimate soiling impact, recommend cleaning and maintenance actions, and move from reactive operations to proactive, data driven solar asset management.

About the Role

We are looking for an Applied Scientist to join our R&D team and build reliable algorithms for solar performance analysis, soiling detection, and maintenance optimization. This role combines physics-guided reasoning, statistics, and machine learning to work with messy real-world plant data and help move models into production pipelines, APIs, dashboards, and customer-facing analytics.

Responsibilities

• Build and maintain robust solar data pipelines for SCADA, inverter, MPPT, irradiance, weather, and operational time-series data, including cleaning, alignment, anomaly detection, and quality control.
• Develop solar performance analytics, soiling detection, energy-loss attribution, forecasting, and maintenance optimization models using physics-guided and data-driven methods.
• Design validation, backtesting, uncertainty scoring, data-quality gates, and model-monitoring workflows to ensure reliable production performance.
• Collaborate with software engineers to deploy model outputs through APIs, batch or streaming pipelines, dashboards, reports, and customer-facing analytics.

Required Skills

• Strong Python and machine-learning skills, including NumPy, pandas, scikit-learn, time-series workflows, model selection, validation, interpretability, and error analysis.
• Ability to combine data-driven modeling with physical reasoning, write clean and documented code, and communicate assumptions, limitations, and modeling decisions clearly.
• Experience with time-series modeling, signal processing, feature extraction, seasonality, change-point detection, anomaly scoring, and uncertainty estimation.
• Ability to work with imperfect telemetry data, including missing values, sensor faults, outliers, timestamp issues, drift, and non-stationary behavior.

Nice to Have

• Experience with solar, renewable energy, weather, satellite, or environmental time-series data.
• Familiarity with PV performance concepts such as irradiance, performance ratio, specific yield, temperature effects, curtailment, clipping, trackers, and solar position.
• Experience with physics-guided ML, hybrid modeling, anomaly detection, fault classification, or predictive maintenance.
• Experience with production ML workflows, including model monitoring, drift detection, retraining, optimization, scheduling, and deep-learning frameworks such as PyTorch.

What We Are Looking For

We are looking for someone who can combine strong data science and machine-learning skills with practical engineering judgment. The ideal candidate is comfortable working with messy real-world solar farm data, understands time-series modeling and anomaly detection, and can use physical reasoning alongside data-driven methods.

This person should be able to build reliable models, explain their assumptions clearly, write clean and maintainable code, and help move algorithms from research notebooks into production-ready analytics.

This is a full-time, in-person role based in Kitchener, Ontario.

As part of the hiring process, selected candidates will complete a take-home technical project in the second round.