Yan Kononenko

Led the build of a multi-tenant SaaS on Next.js (React, TypeScript) with tRPC/REST and PostgreSQL (Prisma, pgvector); shipped core CRM flows in 8 weeks, improving time-to-first-value by 35% for pilot customers. Implemented LLM-powered RAG search using OpenAI/Anthropic, pgvector embeddings, and LangChain/LangGraph; delivering semantic search and entity dedupe/matching that cut recruiter lookup time by 40%. Designed background processing with Inngest/Temporal for idempotent jobs (embedding pipelines, email/SMS sync, webhooks); added retries and DLQs with Redis to reach 99.9% job success. Set up CI/CD with GitHub Actions, per-PR previews on Vercel, feature flags via GrowthBook, and OpenTelemetry traces with Sentry/Datadog; reducing mean time to diagnose by 45%. Built real-time features (activity timelines, presence, notifications) via WebSockets/Server-Sent Events and Redis streams; increasing user engagement by 22%.

Developed React + TypeScript frontends and Node.js/Go APIs with PostgreSQL and Redis caches; improved critical page load time by 28% via query tuning and CDN caching on Cloudflare. Built background jobs (queues, retries, schedules) for enrichment and notifications using BullMQ/Temporal, enabling reliable processing of 10M+ events/month with strict SLAs. Implemented LLM integrations for summarization and recommendations with Vercel AI SDK and streaming UIs, optimizing latency vs. cost through model routing and token budgets. Hardened auth and enterprise SSO with WorkOS/Okta, added fine-grained RBAC, and comprehensive audit trails, supporting multi-tenant compliance requirements. Automated infra with Terraform and Helm across AWS EKS; added GitHub Actions pipelines and preview environments, shortening cycle time from days to hours.

Built data services in Node.js/Go over PostgreSQL with careful relational modeling, transactions, and indexing, reducing heavy-report query times by 60%. Delivered real-time ingestion and event-driven features using Kafka and Redis, supporting downstream analytics and CRM timelines. Added observability (Grafana Cloud/Prometheus/Sentry), cutting P1 detection time by 50% and enabling SLO dashboards. Introduced API versioning (REST/GraphQL), typed contracts, and schema migration practices, stabilizing integrations during rapid iteration. Drove privacy-by-design controls (PII field-level encryption, role scoping, redaction), improving audit outcomes and customer trust.