Filip – AWS, React, Typescript

The company unifies the rental market through a dual-listing architecture that aggregates external partner feeds alongside native landlord listings. Users access the complete rental inventory from a single interface, with clear source attribution for transparency.

The search experience centers on an interactive map powered by PostGIS. Users can draw custom search areas, apply comprehensive filters, and switch between split, list, and map views. An AI concierge assistant provides conversational guidance for users who prefer natural language queries over manual filtering.

For landlords, the platform offers a complete management system: multi-step listing creation, auto-save drafts, drag-and-drop image uploads, and publish controls. The partner integration handles scheduled synchronization, image caching, rate limiting, and compliance tracking automatically.

Being mobile-first with Capacitor for native iOS and Android apps, the platform supports multiple languages with localized formatting. Personalization features—saved searches, favorites, and account dashboards—ensure users can pick up exactly where they left off.

• Architected and built the full-stack platform using Next.js, NestJS, PostgreSQL, and PostGIS
• Designed and implemented the dual-listing aggregation system (external partner feeds + native landlord listings)
• Built scheduled synchronization jobs with image caching, rate limiting, and compliance tracking
• Developed advanced geospatial search using PostGIS, including polygon drawing and bounds-based filtering
• Implemented interactive map experience using MapLibre GL and Mapbox Draw
• Designed comprehensive filtering system with real-time updates and multiple view modes (split, list, map)
• Engineered AI concierge assistant for natural-language property discovery
• Built complete landlord management system (multi-step wizard, drafts, uploads, publishing controls, analytics)
• Implemented authentication with OAuth 2.0 (Google, Apple) and JWT-based session management
• Designed personalization features (saved searches, favorites, notifications, dashboards)
• Developed cross-platform mobile apps using Capacitor for iOS and Android
• Implemented internationalization with localized formatting and multi-language support
• Managed infrastructure (AWS S3, CloudFront, Docker) and optimized performance and scalability
• Owned overall system architecture, UX decisions, and technical roadmap end-to-end

A comprehensive knowledge management platform that uses advanced AI processing to automatically organize, synthesize, and structure conversations into intelligent knowledge bases. The system employs hierarchical categorization, semantic search, and unified knowledge trees to create a living repository of insights.

A Chrome browser extension enables one-click import of ChatGPT conversations with automatic detection, secure authentication, and real-time status notifications. The extension extracts conversation data from ChatGPT’s internal structures, supporting both new imports and incremental updates to existing conversations.

• Architected and built the full-stack platform using Next.js, Supabase, and TypeScript
• Designed and implemented the AI processing pipeline using GPT-4o-mini, GPT-4.1-nano, and GPT-5-mini
• Built a two-stage parallel processing system that improved performance by ~3x on complex conversations
• Implemented semantic search using vector embeddings and cosine similarity
• Developed LLM-powered knowledge tree merging with smart deduplication logic
• Built a Chrome Extension (Manifest V3) for one-click ChatGPT import with incremental updates
• Implemented end-to-end encryption with user-specific master keys
• Developed interactive knowledge tree visualization using React Flow and D3.js
• Configured Supabase with row-level security and workspace isolation
• Implemented type-safe validation using Zod
• Led product architecture, UX decisions, and overall technical direction

This digital platform was developed to support spatial data visualization and management for a regional environmental initiative. It features a public-facing website and an administrative panel that enable users to georeference actions, interact with dynamic maps, and manage territorial information. The platform leverages technologies such as Leaflet and OpenStreetMap to deliver an intuitive interface for visualizing geographic data and coordinating activities within the protected area.

Main responsibilities included the following:

• developed the interactive map interface using Leaflet integrated with OpenStreetMap data;
• built and maintained the admin panel for spatial data management and content updates;
• designed backend APIs with Node.js and Express, storing georeferenced data in MongoDB;
• ensured responsive UI with React and Tailwind CSS for both public and admin portals;
• collaborated closely with stakeholders to tailor geospatial functionalities to user needs.

This digital platform was developed to support Bergamo’s participation in the EU Climate City Contract. It includes a public-facing portal for communication, a secure stakeholder module designed for municipal workflows, and integrated AI features for storytelling and image generation to make climate data and initiatives more engaging and accessible. A built-in newsletter system enables continuous updates to the public and stakeholders. The platform was customized in close collaboration with Bergamo’s Environment Department and over 40 local entities. It is designed to be scalable, modular, and extensible, with the capacity to support millions of potential users across cities and regions. Overall, the platform strengthens governance, streamlines operations, and fosters transparency and collaboration among diverse stakeholders working toward net-zero goals.

Main responsibilities were:

• architected and developed modular monolith and microservices platforms across multiple projects using Node.js, Express, MongoDB, React (Next.js + Tailwind);
• led the development of sustainability-focused tools, including a content management system using AI for copy generation and tagging (OpenAI APIs);
• built and maintained several internal tools, including an admin panel and newsletter delivery system using Vercel, AWS S3, and MongoDB Atlas;
• designed scalable backend APIs, frontend components, and applied DDD, TDD for robust code quality across various customer-facing applications;
• set up infrastructure, CI/CD pipelines, and environments (Elastic Beanstalk, SSH, GitHub Actions);
• mentored junior developers and established Agile workflows (Jira, GitHub);
• refactored legacy modules to follow SOLID/DRY principles, improving maintainability and scalability.

The Italian Judiciary Legal Data Platform is a secure government-grade application designed to process, organize, and display large volumes of legal case data used across Italy’s court system. It enables court clerks, attorneys, and judicial staff to efficiently navigate case files, procedural updates, and historical records.

Built with Astro as the primary framework, the platform is optimized for long-term stability and extremely high performance.

Main Features:

• High-speed browsing of bulk legal case data, indexed and searchable
• Secure, role-based dashboards for judiciary staff and attorneys
• Interactive components (via Astro Islands), including:- Case timeline visualizations
• Document metadata explorers
• Large dataset filtering and sorters
• Bulk document ingestion & automated indexing
• Cross-referenced legal codes and procedural documentation
• Audit logs and compliance-oriented access tracking
• Edge deployment with Cloudflare Workers for low-latency access

• Architected the front-end system using Astro, combining SSG for public documentation with SSR for secure internal dashboards;
• Developed data-heavy interactive modules using Astro Islands to ensure performance while enabling real-time interactions;
• Implemented high-performance search, filtering, and table components capable of handling very large legal datasets;
• Built case timeline views, metadata inspectors, and procedural tree components as isolated interactive islands;
• Designed role-based routing and layout structures for clerks, attorneys, and judiciary staff;
• Integrated front-end with ElasticSearch and MySQL for fast legal case lookup and data retrieval;
• Implemented a bulk document ingestion interface, supporting automated indexing and classification;
• Optimized page speed and Core Web Vitals, consistently achieving 95–100 performance scores;
• Leveraged Astro’s zero-JS rendering to improve reliability on government systems and low-power devices;
• Collaborated closely with backend and data teams to align UI workflows with legal processes and compliance requirements;
• Set up secure CI/CD pipelines using GitHub Actions with Cloudflare Workers for automated deployments.

This project was an automation platform built to help small businesses streamline their social media marketing. It scraped business pages and used AI to automatically generate posting schedules and content plans, enabling efficient and consistent engagement across social platforms.

Among others, Filip carried out the following:

• built backend services in Go and frontend components with Next.js and related frameworks;
• developed AI-driven scraping and scheduling algorithms to automate content posting;
• ensured scalability and responsiveness of the platform for small business users.

A semantic tagging service designed to enhance the classification and discovery of sustainability-related content. The system used natural language processing to automatically generate tagging matrices, improving content organization and searchability across environmental initiatives and databases.

Scope of duties included, but was not limited to:

• designed and implemented semantic tagging algorithms using NLP tools;
• built scalable REST APIs with FastAPI for content tagging and retrieval;
• collaborated with stakeholders to refine tagging criteria aligned with project goals.