资深首席人工智能工程师
Senior Principal AI Engineer
职位描述：
- 作为商业人工智能卓越中心（CAI CoE）中最资深的个人贡献者工程师和主要技术权威，负责企业级AI即服务（AIaaS）的技术愿景。
- 作为全栈AI工程师，精通整个生命周期——数据、模型训练与微调、检索、编排、评估和生产运维，能够深入任何专业领域，并将各个部分整合成连贯且具备生产级别的系统。
- 此职位需要同时具备传统AI/ML、生成式AI产品工程和传统软件架构的技能。
- 建立代理编排、工具和MCP设计、检索、模型训练以及负责任AI的企业参考架构、铺路模式和技术标准。
- 领导最复杂、最模糊的跨团队项目，跨越多个价值流，并审查和影响产品团队的AI设计，以确保符合企业标准。
- 制定长期技术方向，同时深入参与最关键的人工智能基础设施和服务。
主要职责
· 设定PDD范围内AIaaS的多年技术愿景、参考架构和企业标准，并作为组织中最具挑战性AI问题的最终技术权威和升级点。
- 定义并负责传统AI/ML和大语言模型（LLMs）的企业模型训练策略，并在问题需要时亲自进行模型训练、微调（例如QLoRA、LoRA、PEFT和全量微调）和评估。
- 建立从商业AI产品中获取、清洗、版本化、存储和管理训练数据的标准（包括数据溯源、许可/同意和PII），并构建其背后的大规模数据和特征管道。
- 架构中央AI系统的编排和抽象层，连接大语言模型（LLMs）与工具、数据和子代理，并制定MCP服务器、工具界面设计（每个LLM应暴露的工具数量和每个服务器的API数量）以及何时使用专用子代理而非直接暴露工具的标准。
- 端到端设计检索/RAG系统——分块策略、嵌入、向量存储、混合搜索、重新排序、上下文组装和记忆。
- 负责企业AI系统的评估、可观测性和安全性策略，包括离线和在线评估。
查看英文原文
Job Description:
- Serve as the most senior individual-contributor engineer and principal technical authority within the Commercial AI Center of Excellence (CAI CoE), owning the technical vision for AI-as-a-Service (AIaaS) enablement at enterprise scale.
- Operate as a full-spectrum AI engineer fluent across the entire lifecycle — data, model training and fine-tuning, retrieval, orchestration, evaluation, and production operations — able to go as deep as any specialist and to compose the pieces into coherent, production-grade systems.
- This role requires both traditional AI/ML, GenAI product engineering, and traditional software architecture skillsets.
- Establish reference architectures, paved-road patterns, and enterprise technical standards for agentic orchestration, tool and MCP design, retrieval, model training, and responsible AI across production systems.
- Lead the most complex, ambiguous, cross-team initiatives spanning multiple value streams, and review and influence AI designs across product teams to ensure alignment with enterprise standards.
- Set long-term technical direction while remaining deeply hands-on with the most critical AI infrastructure and services.
Key Responsibilities
· Set the multi-year technical vision, reference architectures, and enterprise standards for AIaaS across PDD, and act as the final technical authority and escalation point for the organization's hardest AI problems.
- Define and own the enterprise model-training strategy across traditional AI/ML and LLMs, and personally train, fine-tune (e.g., QLoRA, LoRA, PEFT, and full fine-tuning), and evaluate models when the problem demands it.
- Establish standards for how and where training data from Commercial AI products is sourced, cleaned, versioned, stored, and governed (lineage, licensing/consent, and PII), and architect the large-scale data and feature pipelines behind them.
- Architect the orchestration and abstraction layers of the central AI system that connect LLMs to tools, data, and sub-agents, and set standards for MCP servers, tool-surface design (optimal number of tools exposed per LLM and APIs per server), and when to use specialized sub-agents versus direct tool exposure.
- Design retrieval/RAG systems end to end — chunking strategies, embeddings, vector stores, hybrid search, re-ranking, context assembly, and memory.
- Own the enterprise evaluation, observability, and safety strategy for AI systems, including offline and online evaluation, tracing, red-teaming, guardrails, and responsible-AI and compliance requirements.
- Drive build-versus-buy, model and vendor selection, and long-term architectural bets, anticipating where the field is heading and preparing the organization to adopt it.
- Optimize the performance, cost (token and inference economics), scalability, and reliability of AI workloads in partnership with Security, Cloud Platform, and SRE teams.
- Multiply the organization: mentor and grow Principal and Staff engineers and raise the AI engineering bar.
Required Qualifications
· 15+ years in AI/ML software engineering with demonstrated Senior Principal-level (or equivalent) impact delivering production AI at enterprise scale.
- Full-lifecycle, hands-on mastery across both specialist domains: (a) training and fine-tuning traditional AI/ML models and LLMs — including parameter-efficient methods (QLoRA/LoRA/PEFT), quantization, distributed training, and rigorous evaluation; and (b) LLM orchestration, agentic systems, tool/MCP design, and retrieval/RAG in production.
- Deep expertise in distributed systems, cloud-native architecture, and large-scale data and feature pipelines.
- Strong command of data management and governance: dataset storage architecture, versioning, lineage, quality, PII handling, and licensing/consent for training data.
- Proven ability to design developer platforms, APIs, reusable SDKs, MCP servers, and multi-agent orchestrations that many teams depend on.
- Rigorous, demonstrated approach to AI evaluation, observability and tracing, and responsible-AI guardrails.
- Expertise with cloud platforms (Azure strongly preferred) and a track record of optimizing AI workload cost, performance, scalability, and reliability.
- Demonstrated ability to set technical strategy and influence decisions across many teams without direct authority, and to mentor Principal- and Staff-level engineers.
- Experience operating in regulated SaaS environments and meeting security and compliance requirements.
Preferred Qualifications
· Bachelor's degree in Computer Science, Engineering, or a related discipline; advanced degree preferred. An equivalent combination of education, training, and relevant professional experience is accepted in lieu of a formal degree.
- Industry experience in tax or other regulated domains (insurance, fintech, or healthcare).
- Experience with vector databases and retrieval optimization at scale.
- FinOps for GenAI: experience modeling and optimizing LLM token and inference costs.
- Data science or classical AI background beyond prompt engineering (statistics, feature engineering, and model evaluation).
- Contributions to open-source AI tooling, published research, patents, or recognized technical thought leadership.
- Strong executive communication and technical storytelling skills.
Disclaimer:
The above statements describe the general nature and level of work performed in this role. Other duties may be assigned.
Other Qualifications:
The Winning Way behaviors that all Vertex employees need in order to meet the expectations of each other, our customers, and our partners.
• Communicate with Clarity - Be clear, concise and actionable. Be relentlessly constructive. Seek and provide meaningful feedback.
• Act with Urgency - Adopt an agile mentality - frequent iterations, improved speed, resilience. 80/20 rule – better is the enemy of done. Don’t spend hours when minutes are enough.
• Work with Purpose - Exhibit a “We Can” mindset. Results outweigh effort. Everyone understands how their role contributes. Set aside personal objectives for team results.
• Drive to Decision - Cut the swirl with defined deadlines and decision points. Be clear on individual accountability and decision authority. Guided by a commitment to and accountability for customer outcomes.
• Own the Outcome - Defined milestones, commitments and intended results. Assess your work in context, if you’re unsure, ask. Demonstrate unwavering support for decisions.
Comments:
The above statements are intended to describe the general nature and level of work being performed by individuals in this position. Other functions may be assigned, and management retains the right to add or change the duties at any time.
Pay Transparency Statement:
US Base Salary Range: $229,800.00 - $298,700.00Base pay offered to new hires may vary based upon factors including relevant industry and job-related skills and experience, geographic location, and business needs.* The range displayed does not encompass the full potential of the role, which allows for further growth and career progression.
In addition, as a part of our total compensation package, this role may be eligible for the Vertex Bonus Plan (VOB), a role-specific sales commission/bonus, and/or equity grants.
Learn more about Life at Vertex and connect with your recruiter for more details regarding Vertex's compensation and benefit programs.
*In no case will your pay fall below applicable local minimum wage requirements.
Originally posted on Himalayas