AI & Full-Stack Developer for Energy Efficiency SaaS

AI & Full-Stack Development for an AI-Powered Energy Efficiency SaaS Platform About the Project We are looking for experienced AI / Machine Learning, Computer Vision, Full-Stack, Cloud and API developers , either individual freelancers or development teams, to support the further development of an existing AI-powered platform for energy performance certificates, building data processing and renovation planning. The project is being developed by TB Technisches Büro Otmar Pribitzer GmbH, an Austrian engineering company specializing in energy and building planning. This is not a greenfield project. A functional internally tested prototype already exists, including: * A web application with project-based data collection and database functionality * Initial AI modules for analyzing floor plans, existing energy performance certificates and building photographs * Initial structured data transfer to energy performance certificate software via an XML-based interface * Existing digital workflows for CRM, documents, ordering and billing * More than **10,000 existing project and building datasets** available as a basis for training, testing and quality improvement The objective is to evolve the existing prototype into a **robust, scalable, secure and multi-tenant B2B cloud platform** suitable for external professional users. The planned project start is October 2026. Project Goal The platform is intended to significantly reduce the manual effort involved in collecting, checking and transferring building data for energy performance certificates and renovation planning. The envisioned workflow is: 1. Users upload project documents, floor plans, existing energy certificates, forms and photographs. 2. AI analyzes the available documents, images and plans. 3. Missing, uncertain or contradictory building information is automatically identified. 4. Remaining information is collected during an on-site building inspection through a mobile or responsive interface. 5. Qualified professionals review and approve the extracted data. 6. Approved data is transferred to external energy performance certificate software via API. 7. The same structured building data can later be reused for renovation planning, quantity calculations, cost estimates, economic calculations, subsidies and potentially automated manufacturer quotations. A central principle of the project is Human-in-the-Loop validation. The AI should prepare and evaluate information, indicate confidence levels and highlight uncertainties. Final professional approval must remain with qualified experts. Main Areas of Development The project consists of several modules and work packages. Applicants may propose to deliver the entire project or individual modules, depending on their expertise. 1. Building Data Model & Project Data Management Development of a central structured building data model that consolidates all relevant information required for energy performance certificates, renovation planning and related workflows. Key requirements include: * Project-based storage of plans, certificates, photographs, forms and inspection data * Structured building data model * Data provenance: tracking where each value originated * Distinction between automatically extracted and manually entered data * Confidence and validation status for AI-generated data * Professional approval status * Detection and consolidation of duplicate information * Detection of contradictory or missing data * Auditability of changes and approvals * Synchronization with CRM systems and Microsoft SharePoint where required This is explicitly not intended to become a complete BIM or 3D building model. The focus is a structured and traceable data model containing the information actually required for energy-related workflows. 2. AI-Powered Document and Image Analysis Further development of AI pipelines for extracting structured information from: * Existing energy performance certificates * Technical documentation * Building photographs * Equipment labels and technical plates * Floor plans and scanned plans The system should combine multiple approaches where appropriate rather than relying blindly on a single AI model. Outputs should include: * Extracted structured values * Confidence scores * Validation status * Identification of uncertain results * Identification of results requiring manual verification Possible technologies may include OCR, computer vision, multimodal AI models, machine learning pipelines and rule-based validation. We are open to different technical approaches and expect applicants to justify their proposed architecture. 3. Floor Plan & Building Component Recognition One important component is the extraction of energy-relevant information from floor plans. Relevant elements may include: * Walls * Windows * Doors * Building boundaries * Dimensions * Areas * Orientation * Other energy-relevant building components and geometries The solution must account for different: * Drawing qualities * Scales * Scan qualities * Plan formats * Architectural representation styles The objective is not to generate a complete 3D model. Only information required for downstream energy performance certificate workflows should be extracted and visually identifiable or verifiable. Existing labelled and professionally reviewed data may be used to further train and improve the models. 4. Completeness & Plausibility Checking The platform should evaluate the project as a whole and identify: * Required information that is available * Missing information * Contradictory information * Uncertain AI results * Data that still requires professional verification Both rule-based and AI-supported validation may be used. The system should dynamically generate a task list showing which information still needs to be collected or verified. Corrections made by qualified professionals should be incorporated into a controlled Human-in-the-Loop improvement process. 5. Mobile / On-Site Building Inspection A mobile or responsive interface should support building inspections on site. Functions may include: * Capturing photographs * Recording measurements * Recording building components * Recording technical systems * Assigning information to the correct building, floor, room or component * Synchronization with the central database The interface should guide the user specifically toward information that was identified as missing or uncertain during the previous project analysis. Offline or unstable internet connection scenarios should also be considered in the proposed architecture. 6. Multi-Tenant Cloud Platform The existing prototype must be developed toward a production-ready B2B SaaS platform. Important requirements include: * Multi-tenancy * Separation of company and project data * User roles and permissions * Secure authentication and authorization * Logging and audit trails * Secure data transfer * EU-compliant data storage * Privacy-by-design * Defined deletion processes * Staging, testing and production environments * Reproducible deployment processes * Maintainable and modular architecture * Avoidance of unnecessary vendor lock-in The architecture should allow additional external software integrations in the future. 7. API Development & Integration The existing file-based XML integration with energy performance certificate software should evolve toward a cloud-enabled, versioned API or adapter architecture. The first reference integration will be with an existing external energy performance certificate software partner. The integration may involve: * Mapping several hundred possible parameters * Data validation * Synchronization * Status handling * Error messages * Corrections and updates * API versioning * Integration documentation * Automated API and regression tests The external software provider remains responsible for its own components. The new core platform should remain technically separated. 8. Renovation Planning & Economic Analysis Later modules are intended to use approved building data for: * Renovation alternatives * Step-by-step renovation roadmaps * Quantity calculations * Pre-filled specifications for selected building components * Cost estimates * Energy savings * Cost savings * Economic calculations * Payback periods * Identification of potentially applicable subsidies based on predefined rules and data sources A pilot module for automated manufacturer quotations for selected product categories may also be considered. Technical suitability and commercial placement must remain clearly separated. Expected Deliverables Depending on the modules offered, expected deliverables may include: * Technical architecture documentation * Building data model documentation * Interface and API documentation * Roles and permissions concept * Security concept * Production-quality source code * Source code stored in a repository controlled by the client * Reproducible build and deployment instructions * AI training, validation and testing pipelines * Documented dataset versions * Documented AI quality metrics * Multi-tenant web/cloud application * Mobile inspection interface * Versioned API / adapter implementation * Automated unit tests * Integration tests * API tests * Regression tests * Load tests where appropriate * Security tests where appropriate * Technical documentation * User / operational documentation * Knowledge transfer to the internal development team Acceptance criteria will be defined in more detail during the initial specification phase. For AI-related functionality, measurable quality criteria should be proposed for relevant data classes and document types. AI outputs that are uncertain must remain clearly identifiable and must not automatically be considered professionally approved. Existing Internal Team The selected freelancer or development partner will not work in isolation. The project includes internal expertise covering: * Software development * Cloud development * API development * AI development * Energy performance certificates * Building and energy planning * Domain-specific testing * Data preparation and labeling * Project management The external contractor will collaborate closely with the internal development lead and domain experts through technical discussions, code reviews, integration work and knowledge transfer. Technology Stack The final technology stack has not yet been fixed. We are therefore not looking for someone simply because they use a specific frontend framework or cloud provider. Instead, we expect applicants to propose a suitable architecture and explain: * Which technologies they recommend * Why they recommend them * How the solution would scale * How AI services and models would be integrated * How data security and multi-tenancy would be implemented * How deployment and testing would be organized * Which third-party services or APIs would be required * Expected recurring costs * Potential vendor lock-in * Open-source and licensing considerations Commercially usable open-source and standard components are welcome, but relevant licensing conditions must be disclosed. Who We Are Looking For We are primarily interested in senior / expert-level professionals with proven experience in one or more of the following areas: * Artificial Intelligence * Machine Learning * Computer Vision * OCR / Document AI * Multimodal AI * AI data extraction pipelines * Full-stack web application development * SaaS architecture * Multi-tenant applications * Cloud architecture * API development and integration * Database architecture * Data pipelines * Mobile or responsive web applications * Offline-first / synchronization concepts * Authentication and authorization * Automated testing * DevOps / CI/CD * Software security and privacy-by-design Experience with construction, architecture, energy, PropTech, technical documents, floor plans or engineering software would be highly valuable but is not mandatory if you have strong relevant technical experience. Engagement Model This is a large and technically complex project. We are open to: * Individual senior freelancers * Specialist AI / Computer Vision developers * Full-stack / cloud developers * Small development teams * Specialized agencies You do not need to cover every module. If your expertise is limited to a specific area – for example Computer Vision, AI/data pipelines, cloud architecture, mobile development or API integration – please clearly state which modules you propose to take responsibility for. We explicitly welcome modular proposals. Dependencies on other modules should be clearly identified. What We Need From Your Proposal At this stage, our primary objective is to obtain technical proposals, effort estimates and cost estimates from qualified development partners. Please do not send only a generic introduction. Your proposal should include: 1. Relevant experience Briefly describe projects comparable in technical complexity or scope. 2. Relevant references Provide examples of AI, Computer Vision, SaaS, cloud, API or document-processing projects you have implemented. 3. Proposed scope Clearly state which modules or work packages you would be able to deliver. 4. Technical approach Give us an initial indication of how you would approach the architecture and implementation. 5. Proposed technology stack Explain the main technologies, frameworks, AI models/services and cloud components you would consider using. 6. Project structure Explain how you would divide the work into phases or milestones. 7. Estimated timeline Provide a realistic initial estimate for the modules you propose to deliver. 8. Estimated effort Please estimate effort in hours and/or person-days for the proposed modules where reasonably possible. 9. Pricing Please provide: * Your hourly and/or daily rate * Estimated cost by module where possible * Fixed-price estimates or cost ceilings where reasonably possible * Any assumptions behind your estimate 10. Third-party costs Identify expected recurring or usage-based costs such as: * Cloud hosting * AI APIs * OCR services * Databases * External APIs * Software licenses * Other third-party services 11. Team If applying as a team or agency, identify the people who would actually work on the project and their roles. 12. Availability Please state your earliest possible starting date and expected weekly availability. 13. Risks and open questions Please identify any major technical uncertainties, assumptions or information you would need before providing a final quote. Important Proposal Question To help us identify applicants who have actually reviewed the project, please begin your proposal with: “Energy AI Platform” Then briefly answer the following: Which part of this project do you consider the greatest technical risk, and how would you approach reducing that risk during the first phase? We are particularly interested in the reasoning behind your answer. Development Process We expect an agile development process with: * Short development iterations * Prioritized backlog * Regular demonstrations * Technical reviews * Version-controlled source code * Version-controlled data models and interfaces * Automated testing * Transparent documentation * Continuous knowledge transfer The client should remain able to maintain and further develop the solution without unnecessary long-term dependency on a single external provider. Intellectual Property & Data Newly developed project-specific core components, business logic, project-specific AI workflows, documentation and source code are expected to be transferred to and controlled by the client according to the final contractual agreement. Pre-existing tools, frameworks or proprietary components used by the contractor must be disclosed in advance together with the relevant licensing terms. Project and customer data may only be used for the agreed project purpose. Use of project data for unrelated model training, other customers or other commercial purposes without explicit written authorization is not permitted. Project Timing Planned start: From October 2026 Expected engagement: Long-term / more than 6 months for the overall project However, individual modules may have significantly shorter delivery periods. We prefer realistic estimates over optimistic timelines. Next Step The initial selection will be based on: * Relevant technical experience * Quality of references * Understanding of the project * Proposed technical approach * Ability to explain assumptions and risks * Realistic scope and effort estimation * Communication quality * Commercial transparency Shortlisted applicants will be invited to a technical discussion where the existing prototype, current architecture, available datasets, interfaces and detailed requirements can be presented in greater depth. We look forward to receiving technically well-founded proposals from experienced freelancers and development partners.