Senior Python/AI engineer. I designed and deployed a 19-book, 18-pass multilingual publishing pipeline from zero — running in production today across Polish, Portuguese, English and French. I can build yours next.
Vladimir Sikora · AI Engineer · vova0007@gmail.com
From AI pipeline architecture to production deployment — end to end.
Multi-agent systems with CrewAI, LiteLLM, and custom orchestrators. From prompt engineering to production. I built an 18-pass pipeline processing 19 books across 4 languages.
End-to-end book processing: source audit → multilingual translation → QA → DOCX/EPUB/PDF output → KDP-ready metadata. Built for the Lazarev series (19 books, 1M+ words).
Vector databases, RAG pipelines, knowledge graphs. 4.5M vectors in Qdrant, custom term-memory (MemPalace), automated decision routing between semantic and structured memory.
Python-first integrations with Matrix, Telegram, Gmail, Google Drive, Gemini CLI, ElevenLabs. Custom bridges, systemd services, health monitoring, automatic failover.
LAIRES — automated anti-piracy monitoring across 18+ book platforms (Legimi, Litres, Amazon KDP, Google Play). Blockchain timestamping, auto-reporting, legal evidence packaging.
Language-neutral pipeline cores, per-language prompt stacks, corpus management, kopilochka phrase-mining, Walczak benchmark comparisons. Four active languages, more in profile.
Every book goes through 18 discrete AI stages — from raw Russian source audit to KDP-ready output. Each stage logged, versioned, and replayable.
Interactive demo — click any stage, watch the pipeline process a real book
These aren't demos. Both are running live, serving real clients.
Open to: AI pipeline architecture, translation automation, publishing workflows, RAG/vector systems, multi-agent systems. Upwork · Arc.dev · Direct engagement.