Mining and Materials Science CoPilot

Mining and Materials Science CoPilot submission 👇

💡 Inspiration

Global supply chains for critical minerals like cobalt, nickel, and rare earths face rising ESG, geopolitical, and compliance risks. With 70% of rare earths sourced from China, we were inspired to build an AI co-pilot that brings transparency, sustainability, and accountability to material sourcing—combining AI reasoning, security, and compliance into one platform.

⚙️ What it does

Mining and Materials Science CoPilot scans thousands of ESG and sustainability reports to detect risk patterns such as tariffs, tailings, emissions, tariffs or child labor and maps them to affected supply chains. Using AWS Bedrock embeddings and Claude reasoning, it semantically links mining data to semiconductor and EV industries. It then visualizes ESG clusters and compliance posture through a secure, enterprise-ready UI.

🧠 How we built it • AWS Bedrock (Titan + Claude) for embeddings and reasoning • OpenSearch vector store for document similarity search • Amazon Textract for PDF text extraction • Semgrep for secure code and prompt scanning • Vanta API for compliance posture monitoring • Lovable + Tailwind + shadcn for interactive UI dashboards • Deployed via Vercel + AWS stack

🚧 Challenges we ran into • Handling unstructured ESG PDF formats and incomplete data • Optimizing embeddings for both scientific and policy documents • Integrating security checks (Semgrep) and compliance (Vanta) within a single pipeline • Balancing AI reasoning accuracy with retrieval precision for multilingual sources

🏆 Accomplishments that we’re proud of • Built a secure, compliant AI agent connecting ESG risk analysis with real supply-chain impact • Unified multiple sponsor tools (AWS, Vanta, Semgrep) in one architecture • Created a polished Lovable UI that visualizes ESG clusters and risk scores in real time • Demonstrated practical relevance for semiconductor and EV companies under the CHIPS Act

📚 What we learned

We learned how to combine LLM reasoning, embeddings, and security frameworks to build enterprise-grade AI agents. We also saw the importance of compliance visibility in making AI adoption trustworthy for regulated industries.

🚀 What’s next for Mining and Materials Science CoPilot

We plan to scale the system to millions of ESG and material-science documents, integrate live commodity price and tariff data, and expand to AI-driven material substitution discovery helping industries find sustainable, REE-free alternatives while staying compliant and secure.

Built With

• amazon-web-services
• claude
• cursor
• openai
• semgrep
• vanta