Meeting Optimization for a Financial Networking Conference

The situation

An alternative investment networking platform hosts an annual conference that brings together over 5,000 industry professionals representing more than $10 trillion in institutional assets. The event’s core value proposition is structured one-on-one meetings between asset allocators and investment managers. At the 2025 event, that meant scheduling over 19,000 individual meetings across constrained time slots, limited booth availability, and a venue where walking distances between meeting locations created real timing pressure.

The existing scheduling process could not keep up. Each participant had unique availability windows, meeting priorities based on fund type, and preferences that changed as the event approached. Semi-automated approaches produced schedules with over 10% of requested meetings left unscheduled, and participants spent significant time navigating between distant meeting locations. For an event where every meeting represents a potential multi-million-dollar allocation decision, unscheduled meetings and logistical friction had direct business consequences.

What we built

CorrDyn designed a scheduling optimization engine that treated the problem as a multi-objective constraint satisfaction challenge. The system works in stages. It ingests meeting requests, participant availability, and venue constraints. An initial allocation assigns meetings based on availability and priority ranking. A spatial optimization layer then adjusts assignments to minimize walking distances using venue geometry and booth locations. Iterative optimization passes refine the schedule using genetic algorithm-based hyperparameter tuning and Pareto front analysis to balance competing objectives: schedule density, walking distance, and meeting priority.

The computational infrastructure runs on AWS Batch, which parallelizes execution across high-performance compute instances. This was necessary because the combinatorial complexity of optimizing 19,000+ meetings with individual constraints exceeds what a single machine can process in a reasonable timeframe. The pipeline integrates directly with the platform’s application database, so finalized schedules deploy to the mobile app without manual intervention.

What changed

The optimization engine scheduled all 19,000+ meetings in under 30 minutes. Walking distances between consecutive meetings dropped by 90% compared to unoptimized assignment. Unscheduled meetings fell from over 10% to under 1%, meaning nearly every requested meeting actually happened. Attendees reported noticeably smoother transitions between meetings and higher satisfaction with the overall experience. The system has become a core part of the platform’s event infrastructure, with plans to expand to additional conferences.