IEEE Industry Hackathon 2026
Annual IEEE Southern Alberta YP Industry Hackathon, 48-hour event
2026
Sponsored by ElevenLabs, Databricks, Volaris Group, Hunter Hub, Eudaimonia, and TechConnect Alberta; hosted by IEEE Southern Alberta Section Young Professionals






Project
CityLink: 311 Dispatch Prioritization Agent
Built an agentic dispatch system for Calgary's 311 service, replacing first-in-first-out ticket handling with a priority-scored, auto-replanning scheduler. Using real Open Calgary 311 data, the system scores open tickets by hazard type, wait time, and duplicate reports, assigns them across crews within geographic zones, and automatically re-optimizes when a crew goes offline or a high-priority case is added. An LLM layer interprets plain-language crew updates into structured events for the scheduler, while the underlying prioritization and assignment logic runs on deterministic algorithms. The team placed second overall out of 200+ participants.
Role
- Data pipeline engineering (ticket cleaning, deduplication, filtering)
- Priority scoring and crew assignment algorithm design
- Disruption-handling / re-optimization logic
- Dashboard development (Streamlit)
- Git repository architecture and team coordination
- Testing and sensitivity analysis
Team
Bennett Balogh · Jacob Hailemariam · Semir Haji · Colby Cadden
Technical Focus
We built the pipeline in Python on real City of Calgary 311 open data, using K-means clustering and priority scoring to rank requests and scheduling algorithms to plan the work. The LLM agent runs on the Claude API, the demo was built in Streamlit, and the team shared code through Git.
Key Contributions
- Designed and implemented the ticket deduplication and priority-scoring pipeline, filtering 200 raw tickets down to 99 eligible field-crew jobs
- Built the crew-assignment algorithm using geographic zoning (k-means) to group jobs for each of 8 crews
- Implemented the disruption-replanning logic for when a crew goes offline, including a minimal-perturbation rule that reassigns only affected jobs
- Developed the baseline (oldest-first) comparison to validate results: the agent covered 30 of 30 safety-critical tickets versus 16 under the current method
- Ran a sensitivity analysis across 16 weight variations to confirm the result wasn't dependent on any single tuned parameter
- Built the live Streamlit dashboard showing crew assignments, plan comparisons, and the agent's decision reasoning
- Coordinated the team's GitHub repository structure and resolved integration issues across four parallel workstreams
- Presented the architecture and results in two live judging rounds, placing 2nd overall
Result
Placed 2nd overall out of 200+ participants at the IEEE Southern Alberta YP Industry Hackathon.
