Mindora delivers clinically validated ADHD assessments through an AI-powered platform that helps hospitals reduce wait times, improve diagnostic quality, and expand capacity without additional clinician burden.
At Mindora, I led the end-to-end development of a fully GDPR-compliant platform for an AI-assisted mental healthcare, from ideation through early clinical testing. I built user flows, architected the system design, navigated complex (and emerging) EU regulatory requirements, crafted an Agentic conversational loop to carry the interview, and overcame considerable latency challenges.
User Experience
The project started out with just an idea. I collaborated with the Founder (who is himself a practicing psychiatrist) to discover product requirements and build an intake flow so that we could produce a testable prototype.
Once we had an initial prototype, we were able to get user feedback and iterate on the product direction so that it could become something he was able to robustly test in his practice. Overall, were were able to reduce the time he spent with a patient by 80%, meaning that using the platform, he was able to bring assessment time down to 30-45 minutes, allowing him to see 5x more patients, while maintaining the same level of care.
AI Engineering
Building a low-latency orchestration agent to carry on a real-time video conversation with patients that would at the same time meet all of the the GDPR compliance requirements to protect sensitive personal health information was by far the most challenging technical aspect of this project.
The way in which we managed to solve these challenges was essentially to architect a custom agentic workflow to provide a conversational loop that carefully crafted each and every step within a specific geographical region, and to use PHI filters to ensure secure EU-resident data handling.
The second largest technical challenge was to design multi-layer AI guardrails for input validation, output filtering for PHI, topic boundaries to prevent off-label medical advice, and fallback routing to human clinicians.
Frontend Engineering
Build a React + Next.js platform that integrated with Semble (for patient record management), HeyGen (to provide the digital avatar), Elevenlabs (to provide the voice features), and Azure (for EU LLM orchestration).