SiteLens by Techlife

• 

  Project Poster

Inspiration

Construction failures across Africa—most visibly recent fatal apartment collapses in Nairobi—highlight systemic gaps in safety oversight, skills verification, and data-driven decision-making. At the same time, Africa’s construction sector suffers from fragmented data, limited trust in centralized systems, and decades of unutilized engineering knowledge. SiteLens was inspired by the need to create a secure, collaborative platform where contractors and governments can improve safety, skills, and accountability without compromising data ownership or privacy.

What it does

SiteLens is a privacy-preserving federated learning platform that enables secure collaboration between contractors and government regulators. It allows contractors to participate in federated model training using local construction risk data while retaining full data ownership. Government officials can monitor anonymized safety trends, configure federated learning environments, and support evidence-based policy development. Powered by Gemini 3–generated synthetic data, SiteLens enhances construction safety, delivers personalized, role-specific AI guidance, and supports skills-first talent development across building construction, civil engineering, and architecture.

How we built it

We designed SiteLens as a dual-portal system: a contractor portal for secure data participation, logistics tracking, financial and workforce insights, and a government portal for oversight, policy configuration, and federated learning orchestration. We used Google AI Studio to build and deploy the entire frontend and integrate the backend AI workflows. Federated learning enables decentralized model training without sharing raw data. Gemini 3 is prompted through multiple, role-specific workflows to generate high-fidelity synthetic construction datasets, which the application pre-processes, validates, and stores as structured data for downstream learning and analysis. The system’s data models and long-term roadmap were guided by consultant architects and building construction engineers with over 35 years of industry experience, ensuring practical relevance, regulatory alignment, and real-world scalability while adhering to ethical AI principles.

Challenges we ran into

Balancing regulatory oversight with contractor data privacy was a core challenge. Designing federated learning workflows that are both technically robust and understandable to non-technical stakeholders required careful abstraction. Limited availability of structured construction datasets also necessitated the use of synthetic data, while ensuring realism and relevance. Finally, translating complex AI concepts into practical safety, policy, and skills outcomes demanded strong domain alignment across engineering and governance contexts.

Accomplishments that we're proud of

We successfully demonstrated a working federated learning environment tailored to construction safety and governance. The dual-portal model establishes a new trust framework between contractors and regulators. We integrated Gemini 3 to generate realistic synthetic data and enable rapid risk simulations and AI-driven guidance. We are working with 4 architecture firms and 3 engineering firms to digitize they work which will be used to improve the model and run a proof of concept

What we learned

We learned that privacy-preserving collaboration is essential for AI adoption in regulated industries. Federated learning provides a practical path forward where centralized data is not feasible. Synthetic data can unlock innovation without compromising sensitive information, and role-specific AI guidance is far more effective than generic analytics. We also learned that government–industry collaboration must be designed into the system architecture, not added as an afterthought.

What's next for SiteLens by Techlife

Beyond the hackathon, Techlife aims to evolve SiteLens into a full construction intelligence ecosystem. Next steps include advancing AI-powered BIM rendering and simulations through Gemini 3, developing data-driven construction insurance and financing models, enabling secure licensing and personnel verification, and supporting region-specific policy development. We are also expanding ongoing collaborations with retired engineers and architects to digitize historical knowledge and integrate it into learning models. Ultimately, SiteLens will introduce new AI-driven mathematical models derived from federated learning to optimize safety, sustainability, employment, and infrastructure development across Africa.

Built With

• fastapi
• gcp
• gemini
• google-ai-studio
• postgresql
• python
• typescript