Simulation Model Development and Verification Engineer- Aerospace - Shanghai, China

Simulation Model Development and Verification Engineer- Aerospace - Shanghai, China.

Our Aerospace client offers long term steady work prospects to suitable candidates with experience gained within Simulation Modelling - Development and Verification Engineering. Strongfield have been supporting the civil aircraft programs in China for around 16 years now. If you have interest to live and work in Shanghai, without long hours and high pressure, please forward your CV in the first instance for review.

This is an opportunity for candidates that would like steady long-term work on a new Civil aircraft program, whilst giving them time to see and explore Asia and its culture. Our client has modern facilities and equipment, and they respect the experience that successful candidates can bring to them.

We will work with you throughout the whole process until you arrive and are integrated into Shanghai life.

If Asia has been on your list of continents to see and experience, maybe this could be the opportunity for you. With flights and high-speed transport links, our Engineers have been able to travel and see Macau and Hong Kong for instance.

Job Responsibilities:

Lead the entire optimization process of large-scale model post-training and fine-tuning (architecture design/distributed training/performance tuning), and formulate implementation plans for post-training and fine-tuning.

Build pre-trained base models based on massive industrial data, and explore the integrated application of large-scale models with traditional deep learning and reinforcement learning.

Explore cutting-edge technologies such as RLHF and model distillation, track the latest developments in the field of large-scale model post-training and fine-tuning, and conduct internal team sharing.

Responsible for the architecture design and code development of large-scale model RAG systems, optimizing core modules, and formulating aviation-specific adaptation solutions.

Promote the implementation of RAG technology products, connect with business departments to realize intelligent airworthiness, production scheduling, and other application scenarios, and build a process knowledge graph and LLM collaborative system.

Responsible for the design, development, and optimization of the AI ??Agent system, building and maintaining a private knowledge base, and implementing the Agent and AI Flow framework and process construction.

Complete the integration and deployment cloud/local of multiple types of large models OpenAI Qianwen, etc., and be responsible for the optimization of Python backend services, databases, and retrieval caching technologies.

Led the R&D of core algorithms for multimodal large-scale models, completed the entire training and fine-tuning process, and built a high-quality industry dataset and a dedicated evaluation system.

Researched cutting-edge technologies in multimodal and RAG fields GraphRAG, LLaVA, etc., introduced technical solutions adapted to the aerospace manufacturing industry, and promoted their implementation.

Guided the team in conducting R&D work related to large-scale models, RAG, and AI Agents, completed technical documentation/research report writing, patent applications, and internal team technical sharing.

Promoted the technical integration of large-scale models and the C-Brain AI platform, deeply understood and applied the SIPOC thinking chain methodology, and empowered the intelligent upgrading and process optimization of relevant business scenarios.Job Requirements:

Master's degree or above in Computer Vision, Multimodal Large Modeling, Natural Language Processing, or related fields, with over 8 years of experience in AI-related work.

Experience in successfully optimizing model training algorithms and improving model performance; able to independently undertake algorithm design and development tasks.

Experience in large model-related development, including but not limited to: data cleaning, open-source model fine-tuning, training framework development, reinforcement learning, evaluation, and inference deployment.

Familiar with machine learning principles and algorithms. possesses a deep understanding and practical experience with deep learning algorithms; proficient in Python programming language; familiar with other programming languages ??such as Java. familiar with common frameworks such as PyTorch, Deepspeed, and Megatron. experience in developing large-scale algorithm systems using common deep learning tools such as PyTorch and Tensorflow.

Familiar with various training methods for large language models, including incremental pre-training, instruction fine-tuning, and preference fine-tuning; has practical experience in full-scenario training implementation; understands reinforcement learning techniques PPO/DPO/GRPO/RLHF.

Familiar with full-parameter and LoRA training methods.

Familiar with fine-tuning methods such as QLoRA and Adapter, with practical experience in fine-tuning scenarios.

Familiar with large-scale model training and inference acceleration methods such as DeepSpeed, flash-attn, and VLLM.

Familiar with multimodal domain algorithms, including basic VLM models such as Llava and QwenVL, NLP domain algorithms such as basic LLM models such as BERT, GPT, and Llama, and CV domain algorithms such as basic CV models such as ViT, DINO, SAM, GAN, and Diffusion.

Familiar with data collection, cleaning, and preprocessing processes, with practical experience in constructing training data.

Familiar with large-scale model hint design, proficient in zero-shot and few-shot learning and chain-of-thought inference, and able to optimize hints according to task requirements to reduce model hallucinations.

Candidates must be prepared to live and work in China for 1-3 years initial contract