bukharov.com
Personal projects / April 2026
Goals
- Develop a platform that goes beyond a static portfolio by offering three distinct user journeys—a traditional SPA, an AI-driven conversational agent, and a print-optimized resume—to provide a comprehensive view of my technical expertise and career history.
- Replace traditional, static resumes with a dynamic, project-centric system that highlights high-impact work. By leveraging a centralized headless CMS (Drupal) to feed both the UI and the AI agent, the project ensures all professional data is consistent, up-to-date, and easily accessible.
- Gain hands-on experience in LLM orchestration by implementing a full Retrieval-Augmented Generation (RAG) pipeline. This includes managing a vector store within MySQL and building a FastAPI-based backend to deliver accurate, context-aware responses based on my specific professional data.
- Showcase proficiency in modern web standards and infrastructure by building a fully responsive PWA. The project emphasizes performance and scalability through the use of Server-Side Rendering (SSR), AWS Lambda for serverless compute, and AWS CloudFront for global edge delivery.
This website is a sophisticated, multi-modal personal platform that offers three distinct ways to engage with my professional history. The primary experience is a high-performance Single Page Application (SPA) that provides a traditional, interactive look at my work history and project portfolio. For a more conversational approach, a dedicated chat interface allows visitors to interact with an AI agent that discusses my experience and expertise in real-time. Finally, a specialized print-optimized version automatically formats the site’s content into a professional resume whenever a user initiates a print command from any page. The technical architecture behind the site is a hybrid of modern frontend and backend technologies designed for speed and scalability. The frontend is built using React and TypeScript, ensuring a robust and type-safe user experience. To power the conversational AI, the backend utilizes a FastAPI and Python-based infrastructure implementing a Retrieval-Augmented Generation (RAG) pipeline with MySQL serving as the vector store. This allows the agent to provide accurate, context-aware answers based on my actual career data. Beyond the user-facing features, the platform is engineered as a fully responsive Progressive Web App (PWA) with server-side rendering to optimize performance and SEO. The system utilizes Drupal as a headless CMS, which is uniquely integrated with the FastAPI backend to provide a single, unified admin interface for managing all site content. This ensures that whether information is displayed on a standard page or delivered via the chat agent, it remains consistent and easy to update. The entire ecosystem is built for cloud-native efficiency, with all services running on AWS Lambda for serverless execution. The client-side application is deployed via AWS CloudFront, ensuring low-latency delivery and global availability. By blending these diverse technologies—from headless CMS architecture to serverless compute and AI-driven search—the site serves as both a comprehensive portfolio and a functional demonstration of complex, full-stack engineering.
Outcomes
- Engineered a multi-modal professional platform that successfully integrates a React-based SPA, an AI-driven conversational agent, and a print-optimized resume system into a single, cohesive PWA. This provided a comprehensive, high-performance showcase of my expertise while ensuring a seamless user experience across mobile, desktop, and print formats.
- Implemented a production-grade RAG pipeline using FastAPI, Python, and a MySQL vector store to power a custom AI agent. This outcome demonstrated the ability to bridge the gap between static professional data and dynamic, LLM-driven interactions, allowing for accurate, real-time responses to specific inquiries about my career and technical skills.
- Developed a unified, headless architecture by integrating Drupal as a centralized CMS with a FastAPI backend. This established a single point of administrative control for all site content, ensuring data consistency across both the traditional web pages and the AI’s knowledge base, while utilizing server-side rendering to optimize performance and discoverability.
- Deployed a scalable, cloud-native infrastructure entirely on AWS using Lambda for serverless execution and CloudFront for global content delivery. This resulted in a highly cost-effective and resilient system capable of handling traffic efficiently without the overhead of traditional server management.