Customer Success Services Support (MaaS Service and Model Optimization Direction)

Job Overview

We are seeking a technical expert with deep understanding and practical experience in the field of Model as a Service (MaaS) for a privatization project in Riyadh, Saudi Arabia. You will be based at the client site and serve as the key bridge connecting the client's business needs with the capabilities of large models. Your key responsibilities will be to deeply understand how the client uses large models, diagnose and analyze the performance of the models in specific scenarios, and collaborate with the internal algorithm and product teams to conduct targeted optimizations to ensure the model effect and cost achieve the best balance. Additionally, you will utilize your language skills to handle and optimize model tasks related to Arabic.

Core responsibilities

- Model application analysis and diagnosis: Deeply analyze the performance of the large model in the inference, generation, and review stages within the customer's business scenario, understand and reproduce the bad cases (such as poor performance, high latency, etc.) reported by the customer.

- Model optimization and tuning support: Based on the understanding of the underlying operation logic of the large model, provide customers with optimization suggestions such as Prompt Engineering and Finetune; collaborate with the internal algorithm team to perform model tuning for specific issues (especially those related to Arabic language processing).

- Front-line technical support and consultation: Provide professional technical support and consulting services to customers in areas such as MaaS platform usage, model selection, API invocation, and cost optimization.

- Performance and cost monitoring: Monitor key performance indicators of the customer's model service (such as TPM, TTFT, success rate), analyze resource usage, and provide customers with the most cost-effective model combinations and usage strategies.

- Cross-cultural communication and collaboration: Utilizing language advantages, accurately understand the subtle needs and problem feedback of customers in the Arabic language scenario, and accurately convey them to the internal team to ensure lossless communication. Qualifications for the position

- Experience Requirements: Must have project experience related to large language models (LLM), such as model evaluation, application development, effect optimization, or technical support. Candidates who have worked in MaaS platforms or large AI companies are preferred.

- Technical Skills: - Have a deep understanding of the underlying operation logic of large models, including Transformer architecture, Tokenization, inference process, etc. - Have practical experience in Prompt Engineering and the ability to improve model output quality by optimizing prompts. - Be familiar with model performance evaluation metrics and have experience in model effect evaluation or optimization. - Understand one or more mainstream deep learning frameworks (such as TensorFlow, PyTorch).

- Communication Skills: Excellent communication and expression skills, capable of explaining abstract model concepts and complex optimization strategies in a way that is easy for customers to understand.

Language and Communication

- Chinese: Proficient in the native language to ensure smooth communication with the domestic R&D team.

- English: Fluent in spoken and written English, capable of consulting English technical literature, and participating in the daily work communication of the international team.

- Arabic: Possessing reading and writing skills in Arabic, able to handle and analyze Arabic materials, and understanding the nuances of the local language. This is the core priority for this position.

Work Location and Method

- Work Location: Riyadh, Saudi Arabia.

- Work Method: On-site at the client's location, closely collaborating with the client's data scientists, AI engineers, and business team.

- Duty and Emergency Response: Able to accept 7x24-hour duty and standby arrangements, promptly responding to client's urgent needs and driving problem resolution to a closed loop.

Bonus points:

- Have experience in handling multilingual NLP (Natural Language Processing) tasks, especially in Arabic NLP projects.

- Have practical operation or support experience in model-as-a-service platforms (such as Volcano Ark, Alibaba Cloud Tongyi, Tencent Cloud Hengyuan).

- Have an overseas work or study background, and possess strong cross-cultural collaboration skills.

- Have in-depth research on cost control and performance optimization for model services.

Work Location: In person