depressed intokyo

I ship AI systems end-to-end in production: corporate multi-LLM applications, RAG services, ASR post-processing, model evaluation frameworks, and Gen AI enablement programs at the department level.

At Yandex, I embedded generative AI into product and internal workflows, taking prototypes to daily production use by company employees. I designed and shipped a corporate application with multi-LLM orchestration, using a TypeScript UI with a FastAPI backend and a multi-provider model aggregator.

I also scaled content automation with a RAG service, moving from FAISS to Qdrant with sentence-transformers, and turned evaluation into a data-driven product via an automated AI-model testing framework (chat, image, TTS, transcription, embeddings) with reporting across runs and benchmarks.

Beyond engineering, I ran an internal Gen AI enablement program and mentored designers through Practicum—covering the full generative-AI stack (LLM prompt engineering, agents, ComfyUI-style node graphs, ControlNet, STT/TTS) and lowering the barrier for non-engineering teams.