Johann De La Cruz

Led end-to-end ownership of inventory and HRM modules on a healthcare manufacturing platform, architecting the React UI, designing 14 MongoDB collections, and building 40+ Node.js/Express endpoints. Served as AI integration lead, embedding predictive and generative AI features across the production MERN stack. Diagnosed load-time issues on top-visited views, rewrote 11 MongoDB aggregations with compound indexes, and introduced Redis caching, cutting average page-load time by 52%. Designed and shipped an AI-powered demand-forecasting feature (PyTorch time-series model) served via a Python REST microservice, integrated into the Node.js API, surfacing 7-day stock predictions in a React dashboard, and delivering a 40% reduction in stockouts for high-value SKUs in the first quarter. Implemented a unified auth layer with JWT and 5 RBAC roles across 23 endpoints, aiding HIPAA audit with zero findings. Built a zero-downtime CI/CD pipeline with GitHub Actions and AWS ECS, reducing deployment ti

Core engineer on a 12-person squad delivering a multi-tenant SaaS analytics platform. Owned React/Next.js frontend and NestJS/Node.js backend, implementing subscription billing, licensing, and a real-time WebSocket activity feed. Optimized a 500k+ record admin dashboard by rewriting 6 aggregation stages with $facet/$lookup and adding Redis caching, reducing P95 query time from 8.3s to 0.9s and eliminating the top support ticket for three sprints. Led end-to-end AI integration for churn prediction: fine-tuned a TensorFlow classifier on 18 months of user data, served via a Python inference service, streamed live events to the model, and displayed real-time churn risk scores, cutting at-risk response time by 60% and contributing to a 15% quarterly retention lift. Refactored a monolithic background job runner into a RabbitMQ-based event bus with four NestJS consumers, eliminating timeout spikes and lowering latency by 70%. Containerized eight microservices with Docker and Kubernetes on AWS

Led full-stack development of a custom ERP for a 200-employee manufacturing client. Designed MongoDB schema across 9 core collections, built 30+ Node.js/Express API endpoints, and delivered a React frontend covering inventory, order management, and production scheduling. Enabled onboarding in under two weeks. Resolved a 6-month performance escalation by profiling MySQL queries, adding 8 composite indexes, migrating catalog to MongoDB, and introducing Redis caching on 12 endpoints, reducing checkout load time by 55%. Led AI integration for real-time fraud detection: trained a TensorFlow gradient-boosting model on 2 years of data, deployed via a Python REST inference service, and integrated with the checkout pipeline to surface risk scores with SHAP explanations, cutting manual review time by 35% and reducing chargebacks by 20%. Delivered a telecom client self-service portal in under 10 weeks with a 6-module React SPA and Node.js/Express backend, integrating 4 legacy APIs and implementin