Senior LLM/RAG Engineer

Що очікуємо

• Production experience designing and running RAG pipelines: chunking, embeddings, vector databases, retrieval, reranking, evaluation
• Audio/video processing pipelines: transcription (ASR), speaker diarization, timestamp indexing, handling multi-participant content
• Multi-model / tiered routing architecture — routing lightweight tasks to cheap high-throughput models and escalating complex reasoning tasks to flagship models
• Hands-on evaluation and benchmarking of LLM providers, including Chinese providers (DeepSeek, Alibaba Qwen, ByteDance Doubao, Zhipu GLM, Moonshot Kimi) alongside Western frontier models
• LLM FinOps: token-level cost modelling, unit economics per processing run, and cost projection across a growth curve (10K → 175K → 400K users)