AWS Rekognition Expert Needed to Solve Identical Twin Identity Collision Problem

## Project Overview We have an existing, fully operational digital identity platform with working web and mobile applications, user registration, facial recognition, account creation, and identity verification. Our technology includes AWS Rekognition, AWS Face Liveness, Node.js, MongoDB, PostgreSQL, and AWS infrastructure. Everything is already developed and working. We are looking for a specialist to solve **one specific biometric identity collision problem** within our existing system. This is a development task, not a general consultation or research project. ## The Exact Problem Our system uses facial recognition to identify people and prevent duplicate identities. However, identical twins or individuals with extremely similar faces can be recognized as the same person. For example: 1. Person A is already registered with a unique identity. 2. Person B, who looks almost identical, attempts to register. 3. AWS Rekognition identifies Person B as Person A. 4. Person B reports, "This is not my account." 5. The system must determine whether this is genuinely a different individual or Person A attempting to create a second identity. 6. If they are different individuals, both must receive separate identities and securely access their own accounts. **The biggest challenge is proving that two different people exist without allowing one person to create multiple identities.** ## What Our Existing Developer Has Investigated Our developer has already performed a detailed technical investigation. We have identified the following challenges: - AWS Rekognition similarity scores may not reliably distinguish identical twins. - Increasing facial similarity thresholds can also reject legitimate users. - Capturing additional facial angles does not necessarily solve the problem. - AWS Face Liveness confirms a live presence but does not establish that two similar faces belong to different people. - Using personal PINs can help authenticate established collision identities, but it does not solve the initial problem of confirming two separate individuals. - Simply allowing users to report a collision and create another account introduces duplicate identity and impersonation risks. We have technical documentation and findings that can be shared with suitable candidates. ## Important Technical Restrictions Our solution **cannot depend on**: - National identity cards or government identity databases. - Passports or other external identity documents. - Fingerprint or iris recognition. - Additional biometric hardware or specialized devices. Some of our users may not have any identity documents. We want to work within our existing facial recognition technology, available software capabilities, and application infrastructure. Maximum automation is a major priority, but the solution must not compromise identity security. ## What You Need to Deliver We need you to develop and integrate a practical collision-resolution solution that: 1. Handles situations where two different individuals are recognized as the same person. 2. Provides a technically defensible method for establishing separate identities without enabling duplicate registrations. 3. Securely manages collision groups and account creation. 4. Allows future authentication using facial recognition together with personal PINs or other appropriate software-based credentials. 5. Protects existing accounts, certificates, and personal information. 6. Safely handles failed verification, interrupted sessions, repeated collision attempts, and cases where identity cannot be reliably established. 7. Integrates into our existing backend and application without rebuilding the platform. The solution must fail safely when the available evidence cannot establish a separate identity. Any unavoidable manual exception should be clearly defined and minimized. ## Working With Our Existing Developer Our existing backend developer knows the complete architecture and will coordinate with you throughout the work. He can explain the existing code, provide technical documentation, help with integration, and support testing. **Your primary responsibility will be the collision-resolution algorithm, its security logic, and the working implementation.** We are not looking for someone to take over the entire application. ## Who Should Apply Please apply only if you have strong experience in biometric identity verification, facial recognition algorithms, or similar identity collision problems. We are particularly interested in developers who have dealt with identical twins, biometric false matches, identity deduplication, or secure identity establishment. General experience integrating AWS Rekognition APIs is not sufficient. **We are looking for someone who understands this specific technical challenge and has a credible implementation approach.** We are not hiring someone simply to investigate the issue and prepare another report. ## Budget and Deliverables **Fixed budget: $250 for the complete agreed task.** The expected result is working code integrated into our existing project, with appropriate testing and documentation. Our existing developer will assist with the codebase and integration. ## Questions for Applicants Please answer these three questions: 1. Have you previously solved a similar biometric identity collision or duplicate identity problem? What approach did you use? 2. Given that we cannot use government IDs, identity documents, fingerprints, iris recognition, or additional hardware, how would you approach establishing that two individuals with nearly identical faces are genuinely different people? 3. Can you take responsibility for implementing the solution within our existing Node.js and AWS Rekognition system, working alongside our developer? **Please explain your actual proposed technical approach. Generic proposals, claims that AWS Rekognition alone can reliably distinguish identical twins, or applications offering only an assessment are not what we are looking for.**