diondavid

For Servoy, I built an AI system that understands and interacts with coding documentation to help users with development questions. Using Langchain and vector databases, I gave the LLM real memory so it can retain context and provide accurate, relevant answers across conversations. This setup allows the AI to deeply understand the technical environment and assist developers more effectively.

As the CTO of ScoutsCapital, I spent 1–2 years architecting and building the platform from the ground up with a focus on scalability, performance, and cost-efficiency. I designed and implemented a robust cloud-native infrastructure on Google Cloud’s GKE, combining Kubernetes for orchestration, Grafana for monitoring, Temporal for workflow management, and FastAPI for the backend. The data layer leverages TimescaleDB and PostgreSQL, with Firebase powering real-time interactions. To optimize speed and reduce latency, I integrated caching through both Cloudflare and Google Cloud’s native caching services. The result is a platform capable of achieving the same level of scalability and responsiveness as apps like Instagram, while operating at a fraction of the cost. Beyond that, the system supports more complex workflows and data processing pipelines tailored to investment workflows. I also led development of the Flutter mobile app and Vue web app, ensuring a smooth, consistent experience across devices and platforms.

I created AI models for lead scoring using two main techniques: Random Forests (RF) and Recurrent Neural Networks (RNNs), including LSTM variants. My models—Lead Score (for classification) and Lead Value (for regression)—predict whether a company is a potential customer and estimate the potential revenue. I evaluated and improved these models by rewriting them in Python, measuring performance with metrics like F1-score, AUC-ROC, and MSE, and optimizing for energy efficiency and interpretability using tools like SHAP and LIME.

I built trackyourfiles.com, a file tracking solution designed to monitor and manage digital files, with a focus on email attachments. It allowed users to track when and how recipients interacted with the files they sent, providing valuable insights into client engagement. I led a team of 10–15 developers, making both architectural and business decisions throughout the project.

I’ve spent over a decade building web applications and complex technical systems across a wide range of languages, platforms, and infrastructures. This experience has given me a language-agnostic, architecture-first mindset—focused on core principles like clean design, scalability, performance, and resilience. I apply patterns such as microservices, event-driven architecture, DDD, and CQRS to build modular, maintainable solutions. Beyond full-stack development, I specialize in DevOps and cloud infrastructure. I design CI/CD pipelines, containerize with Docker, and deploy to Kubernetes clusters on GCP, AWS, and Azure. My cloud-native setups include service meshes, observability (Prometheus/Grafana), and cost-optimized, scalable environments with secure access controls. In blockchain, I’ve built decentralized applications in Rust, writing secure, gas-efficient smart contracts for tokenization, governance, and cross-chain systems. I’ve also developed off-chain infrastructure for node orchestration, indexing, and data bridging—grounded in a solid understanding of distributed systems and cryptoeconomics. Overall, I go far beyond traditional full-stack work, owning everything from protocol and application logic to deployment, monitoring, and infrastructure—delivering scalable, secure, and future-ready platforms.