Nikolay Kashev

Modernized a configurable enterprise service-management portal by replacing tightly coupled UIs with reusable React components and independently testable workflow modules. Translated changing client requirements into TypeScript application contracts, keeping business rules explicit and limiting customization-specific logic. Implemented NestJS endpoints backed by PostgreSQL and Prisma to support configurable approval sequences, audit histories, and tenant-specific workflow permissions. Consolidated third-party document and notification integrations behind REST interfaces to isolate vendor-specific failures. Enhanced a Python document classification prototype into a controlled review feature with deterministic validation and operator corrections. Automated Azure App Service deployments through Azure DevOps with Jest tests, environment validation, and repeatable release configurations, and documented architecture decisions and operational handover procedures.

Architected a TypeScript and React operations console for payment exceptions, settlement investigations, and merchant support workflows across independently deployed backend services. Designed Node.js services around transaction lifecycle boundaries, separating read-intensive operational queries from payment-processing workloads. Added Next.js server-rendered views and React Query caching for high-traffic merchant dashboards while preserving transaction data freshness. Integrated banking and payment-provider APIs via typed adapters with explicit timeout policies, idempotent operations, and structured error handling. Orchestrated asynchronous settlement reconciliation using RabbitMQ and PostgreSQL with traceability for late or out-of-order provider notifications. Implemented OpenTelemetry tracing and Datadog dashboards for faster diagnosis of intermittent failures, standardized CI/CD with GitHub Actions checks and Playwright regression tests, and mentored engineers on production debuggi

Built React and TypeScript interfaces for network operations teams, delivering equipment health, alarm histories, and diagnostic timelines from multiple monitoring sources. Developed Node.js aggregation services to normalize inconsistent network telemetry APIs into stable schemas for operational dashboards. Reworked Elasticsearch queries and PostgreSQL data models to keep large alarm histories searchable without overloading upstream systems. Partitioned event processing between RabbitMQ consumers and query-facing services to allow ingestion to continue during dashboard outages. Deployed containerized services on Kubernetes with readiness probes and controlled rollouts, and investigated production latency using Prometheus and Grafana to find expensive aggregation paths and inefficient browser refresh behavior. Expanded Jest and Cypress coverage for alarm filtering, diagnostic navigation, and API compatibility to prevent regressions during platform upgrades, coordinating interface change