EcoTrace AI: Multi-Modal Sustainable Supply Chain Agent

💡 Inspiration

Global trade networks move billions of tons of cargo daily, but the carbon accounting behind these supply chains remains broken, manual, and slow. Most tracking systems rely purely on text auditing, causing severe "alert fatigue" and missed ecological checkpoints.

As a 15-year-old student passionate about planetary preservation, I wondered: What if a logistics network could autonomously "see" emissions and optimize transit tracks before carbon is ever released? This question inspired EcoTrace AI—a multi-modal sustainable agent built to dynamically clean up corporate logistics.

⚙️ What it does

EcoTrace AI acts as an autonomous sustainability coordinator for global enterprise logistics. It takes multi-modal inputs—such as physical cargo photos, multi-language shipping receipts, and live GPS transport logs—and instantly extracts carbon data. The agent monitors transit milestones, flags eco-efficiency violations, and calculates global warming metrics. If a shipping route breaches carbon thresholds, the agent automatically runs optimization models to suggest alternative "green corridors," giving logistics managers immediate, actionable alternatives to bypass traditional supply chain carbon spikes.

🏗️ How I built it

EcoTrace AI is engineered entirely within the Google Cloud and MongoDB ecosystems as a robust, individual developer project:

Vertex AI & Gemini Pro: Handles the multi-modal ingestion pipeline, reasoning through unstructured cargo photos, multi-language shipping receipts, and live route manifests simultaneously.
Google Cloud Agent Builder: Orchestrates the autonomous agentic workflows, parsing multi-step logical constraints to cross-verify cargo weights against vehicle efficiency.
MongoDB Atlas & MongoDB MCP Server: Serves as the high-density transactional storage layer. It maintains a decentralized, secure database ledger of all sustainability scores, operational telemetry logs, and triggers automated eco-compliance alerts instantly when carbon thresholds are breached.

The core carbon math maps the total emissions ($E$) generated across multiple transit legs using the mathematical model:

$$E = \sum_{i=1}^{n} (W_i \times D_i \times C_i)$$

Where $W_i$ represents cargo weight, $D_i$ represents distance traveled in leg $i$, and $C_i$ is the specific carbon intensity coefficient of the chosen transport mode.

🛑 Challenges I ran into

Multi-Language & Messy Data: Foreign suppliers upload receipts in different languages and bad scan formats. I overcame this by using Gemini Pro's native multi-lingual reasoning capabilities to extract structured clean JSON data from messy images.
Predicting Fragmented Logs: Missing maritime transport data often creates gaps. I resolved this by building a contextual predictive logic layer inside the agent to calculate fallback metrics based on regional port-to-port averages.

🎉 Accomplishments that I'm proud of

Successfully designed a zero-lag multi-modal reasoning pipeline that connects visual freight data with carbon tracking math.
Built a scalable, enterprise-grade conceptual agent model as a Class 10 student, proving that complex sustainability tech can be conceptualized by young innovators.
Effectively reduced complex, multi-layered data arrays into clean, actionable "Green Route" alerts that eliminate operational alert fatigue.

🧠 What I learned

I discovered how multi-modal AI agents can transform static data into proactive ecological action. Combining data pipelines with modern cloud telemetry and Model Context Protocols (MCP) proved to me that enterprise software can be both deeply technical and environmentally protective. I also mastered structured prompt engineering frameworks to keep agent actions strictly within sustainable boundary conditions.

🚀 What's next for EcoTrace AI: Multi-Modal Sustainable Supply Chain Agent

The next phase involves integrating live IoT fuel-sensor telemetry directly from ocean cargo fleets into the Vertex AI data layer. I also plan to design localized dashboard extensions for regional customs offices, transforming EcoTrace AI from a corporate logistics tracker into a globally open-source planetary shield for carbon compliance.