EcoMap Cebu: Leading a Hackathon Team From Risky Idea to Working Finalist

Innovation Cup Cebu 2026 asked teams to turn a local problem into a credible digital service under a compressed schedule. Our team, Phishermen, built EcoMap Cebu: a crowdsourced platform that uses image analysis to verify waste reports, maps hotspots across Cebu, and connects environmental reporting to cleanup work. I led the technical direction and built the core path that had to survive the final demonstration.

Choosing a Problem With a Verifiable Outcome

Waste reporting is easy to describe and difficult to operationalize. A photo and location can identify a problem, but a useful system also has to reduce duplicate reports, distinguish severity, show patterns to local government units, and give residents a reason to participate.

We narrowed the first version to one complete loop: a resident captures waste, the system analyzes the image, a verified report appears on the map, and the report contributes to visible cleanup demand. That loop gave every feature a clear role and stopped us from building disconnected screens.

Designing the Technical Spine First

As technical lead, my first task was not choosing every library. It was defining the contracts between capture, analysis, report storage, and map visualization. The frontend needed stable report fields before the backend was finished: coordinates, detected waste type, confidence, severity, image reference, status, and timestamps.

We used sample payloads immediately so interface work could continue while the analysis and persistence path developed. The same payloads later became integration fixtures. This reduced the chance that two working components would fail only when connected near the deadline.

The riskiest path was image-to-map, so I built that before secondary screens. Once a report could move through the complete system and appear at the correct location, the project had a technical foundation rather than a collection of mockups.

Making AI Output Useful

An image classifier by itself does not solve a civic workflow. EcoMap needed to translate model output into decisions people could understand. The analysis view therefore exposes a primary waste type, severity, confidence, and a short recommended action rather than presenting raw detections alone.

Confidence thresholds mattered. Low-confidence results should be reviewed rather than presented as facts, and repeated reports near the same location should strengthen a hotspot without flooding the map with overlapping markers. Even in a prototype, treating model output as evidence rather than certainty made the concept more credible.

Turning Reports Into a City View

The map is where individual reports become operational information. Hotspot markers reveal concentration, while filters let a user move between all waste, plastics, hazardous material, and organic waste. The camera action remains available from the main navigation because reporting and reviewing are parts of the same loop.

We also connected reporting to micro-jobs and rewards. The intent was not to gamify a serious problem superficially, but to show how verified demand could coordinate real cleanup work and recognize participation. That made EcoMap a service model rather than only a complaint channel.

Leading Under a Deadline

The team divided around interfaces: core services and data flow, mobile experience, research, visual identity, and pitch. Short checkpoints focused on integration risks rather than broad status updates. When time became tight, we protected the single demo journey and cut features that did not improve the judges' understanding of it.

I learned that leadership during a hackathon is mostly decision latency. A technically perfect choice made hours late is often worse than a clear, reversible choice made early. Naming owners, freezing contracts, and demonstrating the riskiest behavior first gave everyone space to execute.

Presenting a Working Story

Our final presentation followed one report from capture to analysis to the Cebu hotspot map. The demo used the actual product path rather than a prerecorded simulation. We paired it with a concise explanation of who benefits: residents gain a direct reporting channel, local government units gain geographic visibility, and cleanup workers gain organized demand.

Reaching the finalist stage validated both the concept and the discipline behind the build. The project did not win attention by claiming to solve every environmental problem. It showed one believable system, running end to end, with a path toward broader deployment.

What Comes After the Hackathon

A production EcoMap would need human review queues, stronger duplicate detection, abuse controls, offline report capture, privacy rules for uploaded images, and partnerships with the organizations responsible for response. Model quality would need measurement against local waste categories and real Cebu conditions.

The lasting result is a repeatable approach: define the public outcome, identify the most uncertain technical path, build that path first, and structure the team around stable contracts. Under pressure, scope is not merely what gets removed. It is how the important idea remains visible.