🧠 Built around a reflection engine that delivers context-aware AI insights — not just one-off responses. Real-time streaming chat over WebSockets enables fluid conversations, while a background service builds an evolving user model from journal entries, providing long-term AI context that reasons across time.

🔄 Migrated from LangChain to LiteLLM for a lighter-weight, model-agnostic provider abstraction, reducing coupling while retaining structured memory management and multi-model flexibility. The architecture supports seamless switching between AI providers without tying the system to a single backend.

💰 Production-ready financial management API designed for multi-asset valuation across fiat and cryptocurrencies. Features multi-wallet balances, dynamic exchange rate conversions, and custom transaction classification, backed by high-precision decimal arithmetic (10 decimal places) to eliminate floating-point rounding errors in crypto holdings.

🛡️ Hardened for production with an Nginx reverse proxy enforcing TLS encryption, origin validation, request size constraints, and tiered rate-limiting (strict 1 req/s brute-force protection on auth routes). Backed by comprehensive Pytest E2E test suites and Locust performance benchmarks, fully containerized via Docker Compose on an isolated network.