This poster describes our work of Bayesian optimization of prelithiation process in different scale experiments. Bayesian optimization is a rapidly growing data science technique to find an optimal parameter set with a lower cost. We have evaluated this technique with focusing on a promising process step in a lithium-ion battery production, called prelithiation. Prelithiation is a process step to lithiate anode materials to form solid-electrolyte interface layer before the cell assembly step. Although intensive research has been conducted by many researchers with the lab-scale experiment, a larger scale experiment is important to scaling up the research to the production. To combine different scale experiments, we focus on a technique called multi-fidelity Bayesian optimization. This technique allows researchers to evaluate a low-fidelity information source (accuracy is low, but the cost is also low) many times to reduce the number of trials of a high-fidelity information source (high accuracy with a high cost). To utilise this technique, we have tackled several challenges. Firstly, we have evaluated Bayesian optimization with a small-scale experiment (low-fidelity information source) to materialise our workflow. Secondly, we have developed a research plan to prioritise prelithiation parameters to be evaluated during our project. Although Bayesian optimization reduces the number of trials, the number of process parameters is too huge to be evaluated. Thirdly, we have defined data record templates to track research data through the project. Because a researcher may differ depending on the scale of a conducting experiment, a standardised method to record research data is essential. We will discuss about these practical challenges.