大模型算法工程师（开放域对话）

Nio 2

上海、北京

≈ 24,1 тыс.–35,6 тыс. ₽ · 307–454 $ · 269–398 €

2 083 – 3 083 CNY

🏢 Офис

Без опыта

Стажировка

Продуктовая

Tech

Описание вакансии

Обязанности

•Accelerate inference for low latency deployment

•Apply prompt engineering

•Build end to end dialogue datasets

•Clean and deduplicate raw corpora

•Conduct online A B testing

•Create evaluation datasets for complex interactions

•Develop LLM algorithms for open domain dialogue

•Fine tune foundation models with SFT

•Implement dialogue state tracking

•Improve intent recognition and personalization

•Improve multi turn decision making

•Mitigate hallucinations in tool use

•Optimize models with RLAIF

•Optimize models with RLHF

•Quantize and distill models

•Run offline evaluations

•Train reward models for reinforcement learning

•Use GRPO

Технологии: A/B, A/B Testing, AI Feedback, Agentic RL, B testing, DPO, DeepSpeed, Dialogue State Tracking, Distributed Training, Fine Tuning, Function Calling, GRPO, Human Feedback, Inference acceleration, Knowledge Distillation, Language Models, Large Language Models, Learning from Human Feedback, OpenRLHF, PPO, Prompt engineering, Python, Quantization, React, Reinforcement Learning, Reinforcement Learning from AI Feedback, Reinforcement Learning from Human Feedback, Reward Modeling, State tracking, Supervised Fine Tuning, Tool use, VLLM