AirSci Lab builds data and machine-learning solutions for flight operations, from predicting arrival and departure times to helping teams assess disruption options.
See our projects Contact usAirSci Lab applies data, machine learning and decision-support technology to real operational problems, especially in aviation.
Our projects follow one chain. PETA-Engine predicts when flights will land, take off and be ready to depart. HubTM AI Agentic turns those predictions into options and explains its recommendations. PETA-Project researches better forecasts. Two newer projects, Disruption Packages and Personalised Packages, look after passengers when a flight is disrupted.
Some of this work is in production, some is research and some is in development. The Projects page says which.
Four kinds of work sit behind the projects. Each one is tied to something we build.
Getting flight events, operational tables and passenger data into a shape that models and decision tools can rely on. PETA-Engine writes its predictions to HubTM tables, so the data layer matters as much as the model.
Read more about data strategy and engineeringModels for specific operational quantities. PETA-Engine combines flight-time, taxi-time and turnaround predictions, and PETA-Project tests whether they can be improved.
Read more about machine learning model developmentTurning predictions into one chain of times operators can use: when a flight will land (PETA), take off (PTOT) and when its next rotation will be ready to depart (PETD).
Read more about predictive analytics and forecastingRunning models in production and connecting them to the people who act on them. Predictions are written to HubTM tables, and recommendations reach operators live through Telegram.
Read more about model deployment and integration
Our main language for data manipulation, analysis and machine learning model development.
A deep learning framework for building and deploying neural network models.
A library of machine learning algorithms and tools for preprocessing, modelling and evaluation.
Cloud platforms such as AWS, GCP and Azure for running data and model workloads.
RateHawk is a B2B travel platform that gives travel businesses access to hotel inventory and rates, including through an API. Our partnership brings that hotel supply closer to our work on passenger disruption, where a stranded passenger needs a room in minutes, not hours.
If you work on flight operations or passenger disruption and want to compare notes, get in touch and tell us what you are working on.
Contact us
