Expertise
Technologies and approaches I use in production. Every area is backed by real projects in the portfolio.
Controller-based APIs on .NET 8 and .NET 10 with clear contracts documented via OpenAPI. Centralized error handling through middleware, policy-based authorization and health checks.
Hosted services, Quartz and queue-based jobs with cancellation support, bounded concurrency and retry-friendly synchronization.
Relational databases with EF Core and document models with the MongoDB driver; Repository and Unit of Work patterns, index and query optimization.
Adapter layer that unifies OpenAI, Gemini, OpenRouter and Hugging Face behind one interface. Models and providers change through configuration, avoiding vendor lock-in.
Typed LLM output validated against JSON schemas, system prompts editable from an admin panel and audience-specific prompt variants.
Production-ready assistants with per-session sliding-window memory, persona/expert rules, retries and controlled error responses.
Turning Tesseract OCR output into structured data with LLMs and speech-to-text workflows with Amazon Transcribe.
Logging every AI invocation and keeping latency and cost in check with concurrency limits, caching and retry strategies.
Agentic development with Claude Code and Codex; repository rule files (AGENTS.md/CLAUDE.md), custom skills, and codebase-aware AI workflows with MCP servers (Serena, Context7).
Domain, Application, Infrastructure and API layers with dependencies flowing inward, keeping code testable and extensible.
Instance-per-tenant and domain-based tenant resolution models, per-tenant database isolation and a safe tenant context.
Shared subscription, payment, file and identity services; idempotent operations, distributed locks and cache/database consistency.
Paraşüt, Mikro, Uyumsoft and Xentral integrations with resilient clients using token caching, 429 backoff and status synchronization.
One-off payments, subscriptions and third-party customer/product/price synchronization with Stripe, PayPal, Paddle and iyzico.
Firebase Cloud Messaging, SMTP, SMS and WhatsApp integrations with background-queue delivery.
Multi-service environments with Docker and Docker Compose, scripted production artifact deployments, PM2 and GitHub Actions.
Server provisioning, Nginx/Apache configuration, DNS and domain management, backups, performance tuning and security hardening.
Replica set of primary, secondary and arbiter nodes set up on Linux servers, with the primary running on AWS EC2. When a node goes down, a majority vote triggers an automatic election and the service keeps running; failover scenarios were tested and verified.
Video processing and delivery with S3, MediaConvert and CloudFront; speech-to-text with Amazon Transcribe.