Oleksii Siedov

Designed and built a high-performance Telegram Mini App (TMA) from scratch, bridging the gap between traditional dating and event discovery through advanced AI-driven semantic matching.

Key Technical Achievements:

AI-Driven Matching Engine: Engineered a semantic search core using ChromaDB and sentence-transformers. Unlike basic filters, the system calculates "vibe compatibility" by embedding user bios and event descriptions.

Dynamic Interest Shifting: Implemented a custom vector-blending algorithm. A user's "active vector" shifts based on the events they join, ensuring their feed stays relevant to their current real-world intentions rather than static profile data.

High-Concurrency Backend: Developed a robust FastAPI backend following Clean Architecture principles. Solved race conditions in the matching logic using atomic database-level row locking (SELECT FOR UPDATE), ensuring data consistency under high load.

Seamless TMA UX: Built a responsive, mobile-first frontend with React and Tailwind CSS. Optimized for Telegram’s WebView, featuring a 3-tab navigation system (Profile, People, Events) and non-intrusive notification management.

Performance Optimization: Integrated asynchronous background tasks for heavy AI embedding generation, maintaining a <100ms API response time for the end-user.

•As a Backend Engineer at SkillArena, I focused on building highly scalable and resilient infrastructure for a crypto-gaming ecosystem.

Key Achievements:

Stateless Core Engine: Designed and implemented a fully stateless backend engine using FastAPI, enabling seamless horizontal scaling and high fault tolerance for real-time gaming sessions.

Referral System Architecture: Developed a robust referral and reward system, managing complex logic for user acquisitions, multi-tier rewards, and secure data consistency.

System Integration: Successfully integrated the core engine with external services and crypto-infrastructure, ensuring low-latency communication and 99.9% uptime.

High Performance: Optimized the system for high-concurrency environments, leveraging asynchronous Python to handle rapid spikes in user activity.