Duties
- Model Development & Iteration: Design, train, fine-tune, and iterate machine learning/deep learning models. Provide industry-leading algorithmic solutions tailored to solve complex business pain points.
- End-to-End Algorithm Lifecycle: Take ownership of the full algorithm pipeline, from data cleaning and feature engineering to model training, online evaluation (A/B testing), and final deployment.
- Core Domain Expertise (Select 1-2 based on your team's focus):
- 【Recommendation / Ads / Search】: Develop high-concurrency, low-latency CTR/CVR prediction models (e.g., DeepFM, DIN, Multi-task learning) and optimize recall and ranking mechanisms.
- 【Computer Vision (CV)】: Research and develop algorithms for image classification, object detection, image segmentation, facial recognition, or multimodal generation (AIGC).
- 【Natural Language Processing (NLP)】: Develop algorithms for text classification, sentiment analysis, and information extraction, or lead the fine-tuning (SFT, RLHF) and application of Large Language Models (LLMs).
- Engineering & Performance Optimization: Collaborate with backend and architecture teams to optimize online inference latency and resource consumption, ensuring high availability and stability under massive concurrent traffic.
- SOTA Tracking: Keep abreast of the latest advancements in AI across academia and industry (State-of-the-Art). Conduct technical research and rapid prototyping to validate new ideas against business needs.
Requirements
💡 Preferred Qualifications
High-Performance Deployment: Experience with model compression techniques (quantization, pruning, distillation) and familiarity with high-performance inference and deployment toolchains such as TensorRT, ONNX, or Triton.
- Education: Master’s degree or above in Computer Science, Mathematics, Statistics, Automation, or a related field.
- Programming Skills: Solid foundation in data structures and algorithms; proficient in Python and highly familiar with C++ or Java; excellent coding habits and engineering practices.
- AI Fundamentals: Strong mathematical background (Linear Algebra, Probability, Optimization); deep understanding of classical machine learning algorithms (LR, Tree-based models like XGBoost/LightGBM) and deep learning architectures (CNN, RNN, Transformer).
- Frameworks: Hands-on proficiency with at least one mainstream deep learning framework, such as PyTorch or TensorFlow.
- Comprehensive Skills: Strong data sensitivity and business acumen, with the ability to abstract complex commercial problems into mathematical/algorithmic models; excellent communication skills and ability to thrive in a fast-paced environment.
- Academic / Competitions: Publications in top-tier AI conferences/journals (e.g., NeurIPS, ICML, ICLR, KDD, CVPR, ACL) or top-tier ranking/Gold medals in international data mining/programming competitions (e.g., Kaggle, ACM ICPC).
- Big Data Ecosystem: Proficiency with big data processing platforms such as Hadoop, Spark, Hive, or Flink for massive data analysis and feature mining.
