AI Simulator Uncertainty Engineer: Calibration & Validation

CommonAI Holdings Ltd; Cambridge; Hybrid; GBP 60,000 - 90,000; Full time

Job summary

CommonAI CIC is a non-profit organisation focused on AI systems simulation, uncertainty quantification, and collaboration across AI, HPC, and related fields. The role involves calibrating simulators against measurements and demonstrating the limits of predictions.

You will combine mathematical reasoning with software engineering to calibrate models, test predictions on independent cases, and build uncertainty outputs into the simulation workflow for decision support.

Responsibilities

• Develop and implement methods to calibrate simulator models against measurements and evaluate predictions on cases not used for calibration.
• Distinguish uncertainty from measurements, model assumptions, and parameters, and represent it appropriately.
• Analyse sensitivity and uncertainty in comparisons between alternatives, considering shared sources of uncertainty.
• Develop outputs that show how uncertain a prediction is and when a ranking could change.
• Collaborate with measurement and workload engineers to design experiments that test modeling assumptions and reduce important uncertainty.
• Integrate methods into repeatable simulation workflows and communicate their assumptions and limitations to engineers.
• An example early project may establish a calibration/validation workflow for comparing two architectures with uncertainty outputs.
• Math maturity to reason about models, data, and assumptions behind quantitative claims.
• Software engineering experience to turn numerical methods into readable, reusable tools.
• Experience in quantitative systems performance analysis, statistical modelling, simulation validation, or Bayesian methods would be useful.

Job description

CommonAI CIC is a non-profit organisation focused on AI systems simulation, uncertainty quantification, and collaboration across AI, HPC, and related fields. The role involves calibrating simulators against measurements and demonstrating the limits of predictions.

You will combine mathematical reasoning with software engineering to calibrate models, test predictions on independent cases, and build uncertainty outputs into the simulation workflow for decision support.