Ivan M.

Full Stack Developer with 4+ years building production-grade SaaS platforms, specializing in React, Node.js, Next.js, and NestJS. In the last project I've worked like Tech Lead at Redlizard Studioz, owning the end-to-end architecture of a multi-tenant document management SaaS built from scratch for heavy transport companies designed with per-company subdomain isolation via Cloudflare, ensuring complete data segregation between tenants. I work across the full stack, from PostgreSQL schemas and NestJS microservices to React/Next.js frontends, and I'm comfortable owning infrastructure: deployments on Linux, AWS (Amplify, Elastic Beanstalk), CI/CD pipelines, and third-party integrations. I'm also actively expanding into AI integration, applying production-grade patterns rather than just wrapping a chat endpoint: retrieval-augmented generation (RAG), vector search over embeddings, and tool-use/function-calling agents that execute real actions with backend-enforced permissions and human-in-the-loop confirmation. I'm currently building this an AI assistant for a garbage collection route management platform, where it answers resident questions via RAG over route/schedule data and can create/assign drivers and routes through agentic tool calls, with security and multi-tenant data isolation as first-class requirements. Some things I've shipped: * Architected a multi-tenant SaaS with strict tenant isolation: per-company subdomain routing and data segregation at the infrastructure level, so no tenant can access another's data. * Designed a dynamic billing engine with Stripe Invoices: tier-based pricing algorithm for variable driver counts, deferred charge scheduling, and invoice snapshots in PostgreSQL. * Built a real-time persistent chat (Firestore) with conversation continuity across workflow stages (bid → contract → active work) on a complex multi-role platform. * Engineered a multi-stage email/SMS notification pipeline with scheduler + queue + retry logic (Twilio, Postmark, Redis), handling identity verification and bank authentication flows across multiple countries. * Reduced a critical query from 28s to 3.5s (87% improvement) by redesigning table structure and eliminating N+1 ORM calls. Replicated a full AWS production environment for staging from scratch: Amplify, Elastic Beanstalk, RDS, SSL, DNS routing. * An AI agent layer (Gemini API) for a route-management platform: RAG-based Q&A over operational data, function-calling tools for CRUD actions with backend-enforced permissions, and audit logging for every AI-executed action. Core stack: React · Next.js · Node.js · NestJS · TypeScript · PostgreSQL · TypeORM / Prisma · GraphQL · Redis · Docker · AWS (S3, Cognito, EB, Amplify, RDS, SSL) · Stripe · Firestore · Twilio · Postmark · Keycloak · Cloudflare · Vercel · Railway · CI/CD AI/LLM: Gemini API · LLM Integration · RAG (Retrieval-Augmented Generation) · Vector Databases (pgvector) · Embeddings · Semantic Search · AI Agents · Function Calling / Tool Use · Prompt Engineering · MCP (Model Context Protocol)