Log in
Forgot password

The agenda is open and we are happily accepting your applications. Please send your requests to crew@agifors.org.

Your request should contain

  • Your name and contact email
  • The speaker(s) name(s), their title (as they shall be shown on the agenda) and their contact details (we will not publish them)
  • The desired presentation title and a brief abstract of its content
  • The preferred presentation slot (we try to accommodate as many wishes and preferences as possible, but we trust you understand that this is not always possible)

For technical presentations, the following rules apply:

  • they must not be hidden product showcases, sales pitches or portfolio shows
  • logos, screenshots, product showcases must have a clear and direct technical relation to the content and should be kept to a minimum

In case of any questions and doubts, please reach us via crew@agifors.org.

Monday, 06 April 2026

Start time Details 

Welcome reception 


Tuesday, 07 April 2026

Start time   Details
09:00

Welcome

by Marcel Sol 
09:15

Opening Keynote 
10:00

Technical presentation: The next step in the journey from EI to AI: capturing (useful) feedback from crew.

by Steven Rushworth

At a previous AGIFORS Symposium, I presented findings from my research and discussions with crew communities on the impact of rostering on lifestyle and wellbeing. Building on that foundation, this presentation explores the further practical developments in this area.
It will review recent studies, relevant literature, and industry presentations on the subject. The session will also examine how shifting generational attitudes are reshaping perspectives on work, and the relationship individuals have with their careers.
Most importantly, it will outline and showcase a method for capturing meaningful, dynamic feedback from crew regarding their rosters — and how this feedback can be integrated into the rostering system to drive continuous improvement.
10:30

Coffee break
11:00

Technical presentation: Improving Crew Duty Swap Success Through Predictive Modeling

by Dr. Amine Amrouss

Duty swaps are essential for crew wellbeing, allowing members to adjust schedules to accommodate personal and family commitments. However, the success rate of swap requests remains low due to mandatory roster rules and the need to find eligible and willing counterparts. This presentation introduces a predictive scoring model designed to enhance swap success by analyzing historical behavior, roster preferences, and contextual factors. The model ranks potential swap matches based on predicted willingness, enabling more efficient and successful exchanges. Early results show measurable improvements in swap acceptance rates, highlighting the potential to enhance crew flexibility, satisfaction, and work–life balance.
11:30

Motulus sponsor presentation
11:45

Rois sponsor presentation
12:00

Technical presentation: Quantum Annealing for Airline Crew Trip and Schedule Planning

by Mario Guzzi

Demonstrating a practical path for applying quantum annealing to airline crew planning, showing how realistic operating crew cost estimates can be derived directly from schedules even without a fully completed trip and assignment planning flow. The approach addresses both crew trip construction and crew schedule assignment by sequencing flight segments into feasible trips using explicit operational constraints and trip-start logic, followed by a quantum-based assignment of trips to crews. Both problems are formulated as quadratic unconstrained binary optimization (QUBO) models and executed on D-Wave quantum annealers, with hybrid solvers used for comparison.
12:30

Lunch 
13:30

Jeppesen ForeFlight sponsor presentation
14:00

Technical presentation

by Lana Jansen
14:30

Technical presentation: Autonomous orchestration of third-party optimisation engines for airline crew planning

by Meherzad (Maz) Lakadia

Optimisation engines are widely used in airline crew planning to generate solutions under operational constraints and business priorities. In practice, decision-making rarely relies on a single optimisation run. Analysts execute multiple scenarios to explore trade-offs, assess parameter sensitivity, and justify outcomes for specific operational contexts.
This calibration and scenario management process is typically performed manually by specialised analysts, becoming increasingly time-consuming as operations grow in scale or complexity. Modern engines efficiently solve individual problems, but much of the surrounding exploratory work remains outside the optimiser.
This presentation describes an autonomous orchestration approach applied to a commercial third-party crew planning optimisation engine. The orchestration layer operates externally, treating the optimiser as a black box and interacting only through its standard interface. It autonomously decides which optimisation runs to execute, adjusts parameters, rules, and input data across scenarios within allowable bounds, and determines when sufficient solution quality has been achieved. The system automates multi-scenario sensitivity analysis and trade-off exploration in a single end-to-end workflow.
Although demonstrated for crew planning, the methods are optimiser-agnostic and grounded in established OR and meta-optimisation techniques. The approach produces structured, human-readable insights that reduce manual analyst effort and improve the speed and consistency of optimisation-driven decision-making.
15:00

Coffee break 
15:30

API sponsor presentation
16:00 

AIMS sponsor presentation
16:30

Technical presentation

<RESERVED>
17:00

Technical presentation

<RESERVED> 

Wednesday, 08April 2026

Start time Details 
08:55 Recap day 1
09:00

Technical presentation: Evaluating Systemic Crew Risk and Operational Readiness Across Scheduling, Fatigue, and Crew Logistics

by Daniel Melendez

The future of crew management depends on understanding how scheduling decisions, fatigue exposure, crew accommodation, and operational efficiency interact as a single system. Many organizations optimize these elements independently, unintentionally masking systemic risks that only emerge during disruption or sustained operational pressure.
This presentation introduces a system-level evaluation approach that combines operations research, fatigue risk principles, and AI-based data analysis to assess crew risk and operational readiness end to end—from roster construction and duty sequencing to recovery opportunities, accommodation constraints, and disruption recovery.
The session demonstrates how these methods provide clear visibility into hidden systemic risks, enable a defensible risk and readiness score, and clarify where AI strengthens decision-making versus where it can introduce new risk if misapplied. Attendees will gain a decision-grade view of what degrades if nothing changes, supporting informed choices to ignore, patch, or intervene.
09:30

TA Connections sponsor presentation
09:45

WePlan sponsor presentation

 

10:00

Technical presentation

by Scott Groh
10:30

Coffee break
11:00

Technical presentation: From Combinatorial Search to Sequential Validation: Achieving Real-Time Assignment in Preferential Bidding Systems

by Vigith Kartha, Mangesh Adgaonkar

Traditional airline PBS is NP-hard, requiring slow Branch-and-Price combinatorial search and resulting in opaque, batch-oriented bidding and crew dissatisfaction. This paper proposes a shift to a List-Based, Sequential Validation Assignment. By converting generic rules into a pre-sorted pairing list, the algorithm bypasses the time-consuming Pricing Problem of Column Generation. This breakthrough achieves speeds that make Interactive PBS (IPBS) operationally feasible, enabling continuous feedback. IPBS maximizes transparency, empowers crew to adjust bids, and directly enhances crew satisfaction. This List-Based approach advocates for a new standard in operational efficiency.
11:30

KLM+BCG sponsor presentation 
12:00

get-e sponsor presentation


12:15

Lufthansa Systems sponsor presentation


12:30

Lunch 
14:00 

Social programme 

Thursday, 09 April 2026

Start time  Details 
08:55

Recap day 2 
09:00

Airline update: American Airlines

by Karl Dewitt
09:15

OffBlock sponsor presentation
09:30

Technical presentation: Optimizing Airline Training Planning with Resource-Constrained Modeling

by Dr. Amine Amrouss

Based on projected commercial activity and attrition rates, the Training Planning team must estimate short- and long-term training needs. Each program covering ground, simulator, and flight sessions requires trainers, classrooms, and simulators, each with limited capacity and specific qualifications. We present a mathematical optimization model to enable scalable, proactive scheduling strategies that balance training demand with efficient resource use. The method supports scenario simulation and experimentation, enabling maximized training effectiveness and operational agility in a dynamic airline environment.
10:00

HRS sponsor presentation
10:15

Presentation slot 
10:30

Coffee break 
11:00

Technical presentation: Pilot Line Training Optimization

by Emily Curry, Mathias Lindby

Qualification training for pilots can typically be divided into three stages: ground training, simulators and line flying under supervision (LIFUS). In this presentation we will cover an approach to optimizing scheduling for the LIFUS part of qualification training for both trainees and instructors. Validating this with several airlines we have seen results such as graduating trainees earlier, shorter calendar time to construct LIFUS rosters as well as increasing instructor bid award.
11:30

Technical presentation

<RESERVED>
12:00

Technical presentation

by Scott Groh
12:30

Lunch
13:30

Technical presentation: AI in Crew Planning: Learning from the Past to Improve Future Performance

by Karim Maarouf

Ensuring smooth airline operations begins during the planning phase well before the day of ops, with proactive planning playing a critical role. AI now enables us to harness insights from historical data, revealing what strategies were most effective. In this talk, we’ll share how AI-driven approaches have made crew planning more adaptive and resilient, illustrated by two real-world implementations: one using machine learning to construct efficient pairings, and another applying predictive models to optimize standby crew planning. We’ll also discuss practical lessons learned and key considerations for successful execution.
14:00

Technical presentation: In-Context Learning: Making AI Work for Crew Scheduling

by Viktor Forsman

Large language models have transformed many industries, but their application to crew scheduling remains limited. Due to the domain's complexity—including intricate regulatory frameworks, airline-specific rules, and collective bargaining agreements—general-purpose AI models often produce confident but incorrect answers.
This talk presents a methodology for making LLMs effective within the crew scheduling context. Crew scheduling is a "low-data domain" where specialized knowledge was scarcely available during initial model training. While fine-tuning is often proposed as a solution, it requires large datasets that airlines rarely possess and risks exposing proprietary data. Recent research into in-context learning offers a viable alternative by systematically providing models with domain knowledge at inference time through structured context engineering.
The presentation explores how to structure domain knowledge so that AI models can reason effectively about crew scheduling problems. Central to this approach is systematic context engineering and keeping domain experts in the loop—designing systems that surface knowledge gaps and seek human input rather than hallucinating answers.
Practical applications include ruleset exploration, bidding system support, and optimization outcome explanations. The talk will include demonstrations based on representative crew scheduling scenarios.
14:30

Technical presentation: Mise en Place for the Skies: Delivering Crew the Right Information Before Every Flight

by Dr. Alexander Motzek

Stable airline operations hinge on one consistently underestimated factor: an informed, confident, and empowered crew. "The amuse-bouche is impossible to eat. No spoon." A small thing, fixable mid-flight, if you known. Cabin crew report this kind of detail constantly: catering gaps, route quirks, recurring service issues -- all captured in the feedback systems. Yet on the next flight, the crew needs to re-discover all of it. The loop never closes. With Cosmic AI, we now turn that accumulated knowledge into a personalised pre-flight briefing at Lufthansa Group. Crew take off prepared instead of in the dark.
On the same flight, a *Gold Member in seat 8E quietly discovers that his entertainment screen is dead. The crew knew, technically: "IFE INOP 8A-F" was buried on page 20 of the hold item list. Before the crew can act, the passenger has already called and the service recovery moment is gone. With HIL AI, seat-map intelligence puts every known defect exactly where the crew needs it: visible, immediate, linked directly to the passenger profile. Compensation is one tap away.
Back on the ground, crew members face evacuation training. One instructor, four simultaneous positions, the pressure of being watched, and no way to objectively replay what happened. Disputes are common; burnout is real on both sides. Our AI co-instructor TESSA watches every position at once. It tracks pose, listens to commands, synchronises both against reference procedures, and delivers objective, reproducible feedback. The instructor coaches. The AI observes. Crew walk away with something rare in training: clarity without conflict.
Three flights. Three moments where the right information simply wasn't where it needed to be. This presentation shows how AI is closing those gaps at Lufthansa Group and what comes next.
15:00

Coffee break 
15:30

Closing 
Powered by Wild Apricot Membership Software