Nikolai Zakharov (罗一阳)

Senior Machine Learning Engineer (Multimodal / LLM / RecSys / NLP)

Shanghai, China Tel: +86 136 2105 3870 Email: n.d.zakharov@outlook.com LinkedIn: https://www.linkedin.com/in/nikolai-zakharov-285841a7/ Website: https://yiyang92.github.io/

Download English CV (PDF) →  |  下载中文简历 (PDF) →

Summary

Senior ML Engineer (Multimodal / LLM / RecSys / NLP). 8+ years production AI. Specialized in LLM systems, NLP, recommendation algorithms. Tsinghua MSc. Based in Shanghai. Fluent in Chinese.

Professional Experience

VK (Russia) — Senior Machine Learning Engineer

Russia’s largest social media and technology holding. Jan 2025 – Apr 2026

  • Applied Research, VK AI (Jan 2026 – Apr 2026): Built LLM-based caption generation service for music video recommendations; designed 3-stage distillation pipeline (metadata extractor, audio-text relevance predictor); deployed recsys quality dashboard as key team metric.
  • Video Recommendations, VK Video (Jan 2025 – Dec 2025): Trained and deployed MT5-based content understanding models (metadata extraction, knowledge-base matching); built LLM-based data pipelines on MapReduce/YT; built next-episode recommendation for TV series and similar-items for films. Increased TVT by 25%, reduced duplicates from 60% to 7%.

NIO (Shanghai, China) — Senior Machine Learning Engineer

Mar 2023 – Jan 2025

  • Trained end-to-end transformer planner for trajectory prediction; deployed C++ real-time AD event detection. Applied LLM-based assessment tools to Golden Idea Platform, improving workflow efficiency by 20%.

Tinkoff (T‑Bank) (Moscow, Russia) — Machine Learning Engineer

Russia’s largest digital bank / neobank. Sep 2021 – Feb 2023

  • Trained and deployed Russian Voice Conversion model (95% speaker similarity) serving 10M+ MAU; implemented Go-based streaming inference service and a custom tensor computation library; optimized with TensorRT for cloud deployment.

Huawei Technologies (Shenzhen, China) — Machine Learning Engineer (CBG)

Jul 2019 – Sep 2021

  • Built multilingual trie-based NLU engine (90%+ coverage); trained and deployed RNN-T ASR and voice cloning on device via internal inference framework. Developed gaze estimation model with ARM assembly optimization for on-device deployment.

Research Experience

Kandinsky Lab — Research Project

Dec 2025 – Jan 2026

  • Built prompt optimization system for Kandinsky diffusion model; designed distributed LLM evaluation infrastructure with GPU cluster orchestration.

Huawei Technologies — Research Intern (Noah’s Ark Lab)

2018 – 2019

  • Developed AutoML (DARTS) and GAN-based identity-preserving face recognition augmentation.

Education

Tsinghua University, Beijing, China

  • M.S. in Computer Science (Machine Learning), 2016 – 2019
    • Full Scholarship
    • Thesis: Diversified Image Captioning with Deep Learning (Chinese)
  • Ph.D. in Computer Science, 2022 – 2023
    • Completed coursework; returned to industry

Technical Expertise

  • NLP / LLM: content understanding, NER, entity linking, MT5, ASR, voice cloning, captioning, RAG, model distillation
  • RecSys / Systems: recommendation algorithms, MapReduce, vLLM, TensorRT, C++ real-time, Go streaming, distributed training (FSDP, DeepSpeed), Kubernetes

Publications

中文简历

个人简介

高级算法工程师（大模型/NLP/推荐），8年以上生产级AI经验。专注LLM系统、NLP、推荐算法。清华大学硕士，现居上海。精通俄语、英语、中文。

工作经历

VK（俄罗斯最大社交媒体及科技控股）— 高级算法工程师

2025.01 – 2026.04

  • VK AI / 应用研究组（2026.01 – 2026.04）：端到端构建LLM描述生成服务并落地音乐视频推荐场景，设计三阶段蒸馏管道（元数据提取器、音频-文本相关性预测器）；部署item2item推荐质量看板作为推荐团队核心指标。
  • VK Video / 视频推荐组（2025.01 – 2025.12）：训练并落地多个MT5内容理解模型（元数据提取、知识库匹配）；构建基于LLM的数据处理管道（MapReduce/YT）；实现电视剧下一集推荐管道及电影相似推荐。TVT增长25%，重复结果率60%降至7%。

蔚来汽车 / 自动驾驶规划组 — 高级算法工程师

2023.03 – 2025.01

  • 基于Transformer训练端到端轨迹规划器；部署C++实时AD事件检测系统。将大模型评估能力应用于金点子平台，流程效率提升20%。

Tinkoff银行（俄罗斯最大数字银行）/ 语音技术组 — 算法工程师

2021.09 – 2023.02

  • 训练并落地语音转换系统（Transformer，95%相似度），服务1000万+月活；基于Go实现流式推理服务并开发张量计算库；使用TensorRT优化并部署至云端。

华为 / 小艺语音助手 — 算法工程师

2019.07 – 2021.09

  • 构建多语言Trie树NLU引擎，覆盖90%+语音助手流量；基于自研端侧推理框架，训练并部署RNN-T语音识别与语音克隆至端侧；开发视线估计模型并以ARM汇编优化部署至端侧。

其他经历

Kandinsky Lab / 扩散模型研究 — 研究项目

2025.12 – 2026.01

  • 构建Kandinsky扩散模型的提示词优化系统；设计LLM benchmark分布式评估基础设施：PostgreSQL队列系统与GPU集群编排。

华为诺亚方舟实验室 — 研究实习生

2018 – 2019

  • 开发AutoML系统（基于DARTS），边缘设备精度提升15%；基于GAN的identity-preserving人脸识别数据增强。

教育背景

清华大学，北京，中国

  • 硕士（2016 – 2019），计算机科学与技术（机器学习），全额奖学金 论文：《基于深度学习的多样性图像描述生成》
  • 博士（2022 – 2023），课程完成，返回产业界

技术专长

  • NLP / LLM：内容理解、NER、实体链接、MT5、语音识别、语音克隆、描述生成、RAG、模型蒸馏
  • 推荐 / 系统：推荐算法、MapReduce、vLLM、TensorRT、C++实时推理、Go流式服务、分布式训练（FSDP、DeepSpeed）、Kubernetes

论文发表