14.10.2026 aktualisiert


100 % verfügbar
AI Engineer — LLM Integration, AI Agents, RAG & Voice AI | Full-Stack (Python, TypeScript, Node)
Riga, Lettland
Nur Remote
Riga Technical University - Heat, Gas and Water Technology, 09/2016 - 06/2019 (not completed)Über mich
KI-Softwareentwickler, Full Stack, freiberuflich, remote. Sprachagenten an echten Kundenanrufen, LLM und RAG in vorhandener Geschäftssoftware, KI-Agenten und MCP-Server, Prozessautomatisierung. Kommerzielle Entwicklung seit 2020, KI seit 2023. 50+ Produktiv-Deployments. Allein von Anfang bis Ende.
Skills
AI AgentsLLMJavaPHPAPIsLAMP StackPythonPostgreSQLNode.jsIntegration (Software)BackendFastAPIFull Stack EntwicklungCloudflareFront End
AI / LLM Integration
Daily work with the Claude and OpenAI APIs: server-side integration, streaming, cost/rate control, moderation, and RAG implementations with embeddings, cosine retrieval and cross-encoder reranking. Cloudflare Workers AI.
Full-Stack Web Development
Frontend development with React and Next.js, responsive design-system-driven layouts. Backend with Node.js (Express/Fastify) and Python (FastAPI), REST API design, authentication (JWT, OAuth, sessions).
Databases & Cloud / DevOps
PostgreSQL and SQL: queries, indexes, migrations; ORMs (Prisma, SQLAlchemy). Cloud deployment on Cloudflare Workers/Pages, Git and GitHub, deploy-from-Git, staging/production separation, Docker.
Programming Languages
JavaScript, TypeScript, Python, SQL, PHP, HTML/CSS.
Integrations & Automation
Third-party API integrations (payments incl. Stripe, email/SMS, analytics) and Playwright-based automation and web scraping.
E-Commerce Development
WooCommerce builds end to end: custom themes, product-catalog imports, retail and wholesale pricing, payment and shipping setup.
Erfahrung
Berufserfahrung5 - 10 Jahre
Freelancer seit2020
Sprachen
GermanGrundkenntnisseEnglishverhandlungssicherRussiangutLatvianMuttersprache
Projekthistorie
Delivered dozens of bespoke, per-client web builds and improvements: design-system-driven frontends, responsive across mobile, tablet and desktop, each deployed to Cloudflare with its own domain and DNS, backed by a semi-automated intake and delivery pipeline.
Server-side RAG built solo: overlapping chunking, bge-m3 multilingual embeddings, cosine retrieval, cross-encoder reranking, grounded answers with citations. Cloudflare Workers. Live: https://dynit-rag-demo.jvilcins.workers.dev