{
    "schema_version": "via_ai_product_manager_jobs_v1",
    "data_type": "ai_product_manager_jobs",
    "job": {
        "job_title": "顶级外资-AI Engineer-DL/ML(可选Agent等)",
        "job_level": "管培生",AI生成
        "work_category": "实习",AI生成
        "department": "",
        "location": "香港",
        "work_mode": "现场办公",AI生成
        "job_category": "AI模型算法工程师"AI生成
    },
    "company": {
        "company_name": "某证券公司",
        "company_industry": "证券/期货",
        "company_size": "20-99人",
        "company_website": "",
        "company_description": ""
    },
    "compensation": {
        "salary_range": {
            "min": 60000,
            "max": 90000,
            "currency": "CNY",
            "cycle": "month"
        },
        "benefits": ""
    },
    "requirements": {
        "working_years": 0,
        "degree": "本科",
        "language": "英文",AI生成
        "visa_support": "",
        "responsibilities": "",
        "requirements_text": "有国际期刊/会议论文发表\n英文读写能力良好\nAgent\n业务导向/研究导向\n深度学习\n计算机相关专业\nMATLAB\n深度学习经验\n智能营销/广告\nC/C++\nCNN/RNN/LSTM\n算法\n机器学习\n软件\n智能驾驶\n搜索/推荐\nTensorFlow/PyTorch\n机器学习算法/工程化经验\n语音/图像识别\n云计算\nPython\n\n毕业时间：不限\n\nPrimary Locations: Hong Kong and Shanghai\nProgramme Type: Internship (Off-Cycle and Summer)\nDuration: 3–6 months (Off-Cycle) / 10–11 weeks (Summer)\nEligibility: Open to 2028 and later graduates, as well as 2027 graduates who can start an internship immediately.\nVisa: Hong Kong work visa sponsorship is available.\n About the Role\nA leading global investment firm with decades of track record is expanding its Asia Pacific internship programme. As an AI Engineering Intern, you will collaborate with investment professionals on real investment questions and industries. You will design, prototype, and deploy AI-powered systems that accelerate research, automate workflows, and translate frontier AI capabilities into practical investment research products.\n Responsibilities\n- Collaborate with investment professionals on real investment questions and industries.\n- Develop AI agents and automated workflows for company and industry research.\n- Build multi-agent systems that integrate large language models, tools, APIs, and internal knowledge.\n- Prototype tools integrating internal knowledge, public data, documents, APIs, and analytics.\n- Structure and automate multi-source data to support investment research.\n- Track and test emerging AI models, developer tools, and open-source projects.\n- Translate AI capabilities into practical research products with real applications.\n- Evaluate output accuracy, source quality, usefulness, safety, and interpretability.\n- Refine workflows based on feedback and production performance.\n- Deploy validated AI systems into production research environments.\n Research Directions (Select 1–3)\nDirection 1: AI Agent-Driven Quantitative Research *(First Priority)*\nBuild autonomous decision-making financial intelligent agents by integrating large language models, reinforcement learning, and quantitative strategies. Design multi-agent collaborative frameworks where agents assume roles in data cleaning, signal mining, risk monitoring, and market sentiment analysis. Explore real-time understanding of unstructured information via LLMs, and develop Agent-based meta-learning strategies that enable systems to adaptively adjust model weights and risk exposure in dynamic markets. Focus on Agent behavior interpretability, hallucination suppression, and trading safety boundaries.\nDirection 2: Machine Learning / Deep Learning Quantitative *(Second Priority)*\nApply ML models to uncover nonlinear patterns using gradient boosting trees, neural networks, Transformer architectures, and more. Conduct rigorous feature engineering, regularization, out-of-sample validation, and model decay monitoring. Experience with XGBoost/LightGBM, scikit-learn, PyTorch/TensorFlow is highly valued.\nDirection 3: LLM & NLP for Investment Research\nDevelop large language model and natural language processing systems for investment research. Build pipelines for document parsing, summarization, information extraction, sentiment analysis, and research report generation. Explore retrieval-augmented generation and domain adaptation for financial text.\nDirection 4: Multi-Agent Systems & Workflow Automation\nDesign multi-agent architectures where agents collaborate on data cleaning, signal discovery, risk monitoring, report generation, and execution support. Develop memory mechanisms, tool calling, planning, and coordination protocols for reliable end-to-end research workflows.\nDirection 5: Reinforcement Learning for Trading & Decision Making\nApply reinforcement learning to trading, portfolio allocation, execution, and adaptive decision-making. Build simulation sandboxes, reward designs, and safety constraints for robust learning in dynamic market environments.\nDirection 6: Alternative Data & Unstructured Data Intelligence\nParse text, satellite, sensor, and other unstructured data to extract effective signals. Build event-driven or decision-support models using NLP, computer vision, and multimodal learning. Design event study frameworks and data quality validation.\nDirection 7: AI Infrastructure & Research Tooling\nBuild scalable infrastructure for model training, inference, evaluation, and deployment. Develop internal research tools, data pipelines, experiment tracking, and API integrations for investment research.\nDirection 8: Knowledge Graphs, RAG & Retrieval Systems\nConstruct financial knowledge graphs and retrieval-augmented generation systems. Integrate internal documents, public filings, news, and structured databases into reliable, citation-aware research assistants.\nDirection 9: Model Evaluation, Safety & Interpretability\nDesign evaluation frameworks for accuracy, hallucination, bias, robustness, and safety. Develop interpretability tools and guardrails for AI systems used in investment research and trading support.\nDirection 10: AI-Powered Factor Mining & Automation\nUse AI methods such as genetic programming, symbolic regression, and reinforcement learning to generate and evaluate candidate factors. Build automated factor mining pipelines with strict out-of-sample validation.\nDirection 11: Production AI Systems for Investment Research\nTurn prototypes into production-grade AI systems. Work on latency, reliability, monitoring, security, compliance, and human-in-the-loop workflows for investment research.\nDirection 12: Any Other AI Engineering Subfield of Your Interest\n Ideal Candidate\n- Strong programming and analytical ability with excellent engineering instincts.\n- Currently enrolled in a BA/BS, MSc, or PhD in Computer Science, Data Science, Engineering, Mathematics, Statistics, Artificial Intelligence, or a closely related field.\n- Solid background in machine learning, deep learning, large language models, or multi-agent systems.\n- Strong programming skills in Python; experience with PyTorch/TensorFlow is highly valued.\n- Familiarity with LLM APIs, RAG, vector databases, prompt engineering, and agent frameworks.\n- Experience working with large, unstructured datasets preferred.\n- High intellectual curiosity, proactive mindset, and ability to learn quickly.\n- Professional fluency in English and Mandarin with clear communication skills.\n- Interest in investing and industry research; financial experience is a plus but not required.\n- Prior internship experience in AI, software engineering, or data science preferred but not required.\n Technical Stack Requirements\n- Proficiency in Python for data processing, model implementation, API development, and automation.\n- Experience with machine learning and deep learning frameworks such as PyTorch or TensorFlow.\n- Familiarity with LLM APIs, agent frameworks, RAG pipelines, and vector databases.\n- SQL and common relational databases; familiarity with Linux, Git, Docker.\nBonus Points\n- Experience building AI agents, multi-agent systems, or LLM-powered applications.\n- Experience with reinforcement learning, knowledge graphs, or multimodal models.\n- High-performance computing languages such as C++, Rust, Go, Julia, or OCaml.\n- Ability to complete the full pipeline from data cleaning, model training, evaluation to production deployment.\n- Rigorous evaluation design capability, understanding of hallucination control, safety boundaries, and human-in-the-loop validation.\n What We Offer\n1. Industry-leading compensation — base salary paired with a discretionary performance bonus and a comprehensive benefits package, benchmarked against the highest standards of global investment firms.\n2. Work-life balance — protected free space during work hours for deep learning, exploration, focused thinking, or rest and recovery.\n3. Immersive development — a structured training programme covering financial markets, technical skills, and business fundamentals, complemented by individualized development plans.\n4. Mentorship at every level — paired with dedicated mentors and buddies, with direct access to senior leaders and some of the industry’s most experienced professionals.\n5. AI-augmented research environment — access to sophisticated AI tools and resources, with encouragement to apply AI throughout daily research and development workflows.\n6. Global career pathway — high-performing interns are considered for analyst opportunities across investment teams following graduation, with exposure spanning teams and geographies.\n7. Premier workplace amenities — comfortable rest areas, in-house coffee and beverage service, employee fitness facilities, ergonomic workstations, and high-resolution multi-screen setups.\n8. Comprehensive benefits — paid vacation for full-time employees, health and wellness coverage, meal and housing subsidies where applicable, holiday and birthday gifts, and anniversary commemorations.\n9. Collaborative, meritocratic culture — an entrepreneurial mindset combined with institutional strength, where success is driven by performance and independent thinking is valued.\n10. Employee networks and community — inclusive employee resource groups supporting development and connection across backgrounds and experiences.\nNote: Some new graduates may need to complete an internship assessment (not exceeding 3 months) to receive a full-time offer / return offer.\n认证资质\n人力资源服务许可证\n其它信息\n所属部门：Quant or AI Lab",
        "skills": [AI生成
            "模型训练与算法优化",
            "模型微调与对齐",
            "编程与代码实现",
            "上下文/知识工程",
            "Agent系统构建"
        ],
        "preferred_qualifications": "",
        "job_highlights": ""
    },
    "contact": {
        "recruiter_name": "",
        "contact_methods": [
            {
                "type": "apply_link",
                "value": "https://www.zhipin.com/job_detail/f1b3aa584017121d0nN-3t21EVBV.html"
            }
        ]
    },
    "posting": {
        "posting_id": "f1b3aa584017121d0nN-3t21EVBV",
        "post_date": "",
        "hire_time": "",
        "update_date": "",
        "job_url": "https://www.zhipin.com/job_detail/f1b3aa584017121d0nN-3t21EVBV.html"
    },
    "fulltext": "有国际期刊/会议论文发表\n英文读写能力良好\nAgent\n业务导向/研究导向\n深度学习\n计算机相关专业\nMATLAB\n深度学习经验\n智能营销/广告\nC/C++\nCNN/RNN/LSTM\n算法\n机器学习\n软件\n智能驾驶\n搜索/推荐\nTensorFlow/PyTorch\n机器学习算法/工程化经验\n语音/图像识别\n云计算\nPython\n\n毕业时间：不限\n\nPrimary Locations: Hong Kong and Shanghai\nProgramme Type: Internship (Off-Cycle and Summer)\nDuration: 3–6 months (Off-Cycle) / 10–11 weeks (Summer)\nEligibility: Open to 2028 and later graduates, as well as 2027 graduates who can start an internship immediately.\nVisa: Hong Kong work visa sponsorship is available.\n About the Role\nA leading global investment firm with decades of track record is expanding its Asia Pacific internship programme. As an AI Engineering Intern, you will collaborate with investment professionals on real investment questions and industries. You will design, prototype, and deploy AI-powered systems that accelerate research, automate workflows, and translate frontier AI capabilities into practical investment research products.\n Responsibilities\n- Collaborate with investment professionals on real investment questions and industries.\n- Develop AI agents and automated workflows for company and industry research.\n- Build multi-agent systems that integrate large language models, tools, APIs, and internal knowledge.\n- Prototype tools integrating internal knowledge, public data, documents, APIs, and analytics.\n- Structure and automate multi-source data to support investment research.\n- Track and test emerging AI models, developer tools, and open-source projects.\n- Translate AI capabilities into practical research products with real applications.\n- Evaluate output accuracy, source quality, usefulness, safety, and interpretability.\n- Refine workflows based on feedback and production performance.\n- Deploy validated AI systems into production research environments.\n Research Directions (Select 1–3)\nDirection 1: AI Agent-Driven Quantitative Research *(First Priority)*\nBuild autonomous decision-making financial intelligent agents by integrating large language models, reinforcement learning, and quantitative strategies. Design multi-agent collaborative frameworks where agents assume roles in data cleaning, signal mining, risk monitoring, and market sentiment analysis. Explore real-time understanding of unstructured information via LLMs, and develop Agent-based meta-learning strategies that enable systems to adaptively adjust model weights and risk exposure in dynamic markets. Focus on Agent behavior interpretability, hallucination suppression, and trading safety boundaries.\nDirection 2: Machine Learning / Deep Learning Quantitative *(Second Priority)*\nApply ML models to uncover nonlinear patterns using gradient boosting trees, neural networks, Transformer architectures, and more. Conduct rigorous feature engineering, regularization, out-of-sample validation, and model decay monitoring. Experience with XGBoost/LightGBM, scikit-learn, PyTorch/TensorFlow is highly valued.\nDirection 3: LLM & NLP for Investment Research\nDevelop large language model and natural language processing systems for investment research. Build pipelines for document parsing, summarization, information extraction, sentiment analysis, and research report generation. Explore retrieval-augmented generation and domain adaptation for financial text.\nDirection 4: Multi-Agent Systems & Workflow Automation\nDesign multi-agent architectures where agents collaborate on data cleaning, signal discovery, risk monitoring, report generation, and execution support. Develop memory mechanisms, tool calling, planning, and coordination protocols for reliable end-to-end research workflows.\nDirection 5: Reinforcement Learning for Trading & Decision Making\nApply reinforcement learning to trading, portfolio allocation, execution, and adaptive decision-making. Build simulation sandboxes, reward designs, and safety constraints for robust learning in dynamic market environments.\nDirection 6: Alternative Data & Unstructured Data Intelligence\nParse text, satellite, sensor, and other unstructured data to extract effective signals. Build event-driven or decision-support models using NLP, computer vision, and multimodal learning. Design event study frameworks and data quality validation.\nDirection 7: AI Infrastructure & Research Tooling\nBuild scalable infrastructure for model training, inference, evaluation, and deployment. Develop internal research tools, data pipelines, experiment tracking, and API integrations for investment research.\nDirection 8: Knowledge Graphs, RAG & Retrieval Systems\nConstruct financial knowledge graphs and retrieval-augmented generation systems. Integrate internal documents, public filings, news, and structured databases into reliable, citation-aware research assistants.\nDirection 9: Model Evaluation, Safety & Interpretability\nDesign evaluation frameworks for accuracy, hallucination, bias, robustness, and safety. Develop interpretability tools and guardrails for AI systems used in investment research and trading support.\nDirection 10: AI-Powered Factor Mining & Automation\nUse AI methods such as genetic programming, symbolic regression, and reinforcement learning to generate and evaluate candidate factors. Build automated factor mining pipelines with strict out-of-sample validation.\nDirection 11: Production AI Systems for Investment Research\nTurn prototypes into production-grade AI systems. Work on latency, reliability, monitoring, security, compliance, and human-in-the-loop workflows for investment research.\nDirection 12: Any Other AI Engineering Subfield of Your Interest\n Ideal Candidate\n- Strong programming and analytical ability with excellent engineering instincts.\n- Currently enrolled in a BA/BS, MSc, or PhD in Computer Science, Data Science, Engineering, Mathematics, Statistics, Artificial Intelligence, or a closely related field.\n- Solid background in machine learning, deep learning, large language models, or multi-agent systems.\n- Strong programming skills in Python; experience with PyTorch/TensorFlow is highly valued.\n- Familiarity with LLM APIs, RAG, vector databases, prompt engineering, and agent frameworks.\n- Experience working with large, unstructured datasets preferred.\n- High intellectual curiosity, proactive mindset, and ability to learn quickly.\n- Professional fluency in English and Mandarin with clear communication skills.\n- Interest in investing and industry research; financial experience is a plus but not required.\n- Prior internship experience in AI, software engineering, or data science preferred but not required.\n Technical Stack Requirements\n- Proficiency in Python for data processing, model implementation, API development, and automation.\n- Experience with machine learning and deep learning frameworks such as PyTorch or TensorFlow.\n- Familiarity with LLM APIs, agent frameworks, RAG pipelines, and vector databases.\n- SQL and common relational databases; familiarity with Linux, Git, Docker.\nBonus Points\n- Experience building AI agents, multi-agent systems, or LLM-powered applications.\n- Experience with reinforcement learning, knowledge graphs, or multimodal models.\n- High-performance computing languages such as C++, Rust, Go, Julia, or OCaml.\n- Ability to complete the full pipeline from data cleaning, model training, evaluation to production deployment.\n- Rigorous evaluation design capability, understanding of hallucination control, safety boundaries, and human-in-the-loop validation.\n What We Offer\n1. Industry-leading compensation — base salary paired with a discretionary performance bonus and a comprehensive benefits package, benchmarked against the highest standards of global investment firms.\n2. Work-life balance — protected free space during work hours for deep learning, exploration, focused thinking, or rest and recovery.\n3. Immersive development — a structured training programme covering financial markets, technical skills, and business fundamentals, complemented by individualized development plans.\n4. Mentorship at every level — paired with dedicated mentors and buddies, with direct access to senior leaders and some of the industry’s most experienced professionals.\n5. AI-augmented research environment — access to sophisticated AI tools and resources, with encouragement to apply AI throughout daily research and development workflows.\n6. Global career pathway — high-performing interns are considered for analyst opportunities across investment teams following graduation, with exposure spanning teams and geographies.\n7. Premier workplace amenities — comfortable rest areas, in-house coffee and beverage service, employee fitness facilities, ergonomic workstations, and high-resolution multi-screen setups.\n8. Comprehensive benefits — paid vacation for full-time employees, health and wellness coverage, meal and housing subsidies where applicable, holiday and birthday gifts, and anniversary commemorations.\n9. Collaborative, meritocratic culture — an entrepreneurial mindset combined with institutional strength, where success is driven by performance and independent thinking is valued.\n10. Employee networks and community — inclusive employee resource groups supporting development and connection across backgrounds and experiences.\nNote: Some new graduates may need to complete an internship assessment (not exceeding 3 months) to receive a full-time offer / return offer.\n认证资质\n人力资源服务许可证\n其它信息\n所属部门：Quant or AI Lab"
}