GenAI 工程师
GenAI Engineer
Tiger Analytics 是 AI 和高级分析咨询领域的全球领导者，赋能财富 1000 强企业解决最棘手的业务挑战。我们致力于推动 AI 的边界，为更美好的明天提供数据驱动的确定性。我们的 6000 多名技术专家和顾问团队遍布五大洲，规模化构建前沿的机器学习和数据解决方案。加入我们，创造卓越的工作，并塑造企业 AI 的未来。
我们正在寻找一位经验丰富的 GenAI 工程师，具备构建、评估和部署先进生成式 AI 系统的扎实实践经验。理想的候选人应具备代理框架、模型微调和强化学习方面的深厚专业知识，同时在实验、可靠性以及幻觉缓解方面有强烈关注，超越提示工程的范畴。
要求
- 为实际应用场景设计、构建和部署端到端的生成式 AI 和代理式 AI 解决方案。
- 使用 LangGraph、MCP（模型上下文协议）和 A2A（代理到代理）通信模式开发和编排多代理工作流。
- 使用监督微调（SFT）、RLHF 和其他高级技术对大型语言模型（LLMs）进行微调，以提高任务性能和对齐度。
- 应用强化学习方法优化代理行为、决策和长周期任务。
- 设计并执行严格的实验框架，包括离线/在线评估、A/B 测试和基于指标的改进。
- 实施可靠的幻觉减少策略，如检索增强、定位、验证层、置信度评分和自我反思机制。
- 与数据工程师、产品经理和平台团队合作，将 GenAI 解决方案集成到生产系统中。
- 监控、评估并持续改进模型性能、可靠性、延迟和成本。
- 跟踪生成式 AI、代理系统和模型对齐领域的最新研究和进展。
所需资格
- 5 年以上软件工程、机器学习或 AI 相关岗位的行业经验。
- 熟练使用 LangGraph 和构建代理工作流。
- 具备 MCP（模型上下文协议）和 A2A（代理到代理）系统设计的实际经验。
- 在微调 LLMs 方面有实际经验，包括监督微调和基于强化学习的方法。
- 对相关概念有扎实的理解和应用能力。
查看英文原文
Tiger Analytics is a global leader in AI and advanced analytics consulting, empowering Fortune 1000 companies to solve their toughest business challenges. We are on a mission to push the boundaries of what AI can do, providing data-driven certainty for a better tomorrow. Our diverse team of over 6,000 technologists and consultants operates across five continents, building cutting-edge ML and data solutions at scale. Join us to do great work and shape the future of enterprise AI.
We are looking for a highly skilled GenAI Engineer with strong hands-on experience in building, evaluating, and deploying advanced Generative AI systems. The ideal candidate will have deep expertise in agentic frameworks, model fine-tuning, and reinforcement learning, along with a strong focus on experimentation, reliability, and hallucination mitigation beyond prompt engineering.
Requirements
- Design, build, and deploy end-to-end Generative AI and agentic AI solutions for real-world use cases.
- Develop and orchestrate multi-agent workflows using LangGraph, MCP (Model Context Protocol), and A2A (Agent-to-Agent) communication patterns.
- Fine-tune large language models (LLMs) using supervised fine-tuning (SFT), RLHF, and other advanced techniques to improve task performance and alignment.
- Apply reinforcement learning approaches to optimize agent behavior, decision-making, and long-horizon tasks.
- Design and execute rigorous experimentation frameworks, including offline/online evaluations, A/B testing, and metric-driven improvements.
- Implement robust strategies for hallucination reduction, such as retrieval augmentation, grounding, validation layers, confidence scoring, and self-reflection mechanisms.
- Collaborate with data engineers, product managers, and platform teams to integrate GenAI solutions into production systems.
- Monitor, evaluate, and continuously improve model performance, reliability, latency, and cost.
- Stay up to date with the latest research and advancements in GenAI, agentic systems, and model alignment.
Required Qualifications
- 5+ years of industry experience in software engineering, machine learning, or AI-focused roles.
- Strong hands-on experience with LangGraph and building agentic workflows.
- Practical experience with MCP (Model Context Protocol) and A2A (Agent-to-Agent) system design.
- Proven experience in fine-tuning LLMs, including supervised fine-tuning and reinforcement learning-based methods.
- Solid understanding and application of reinforcement learning concepts in production or research settings.
- Strong background in experimental design, model evaluation, and statistical analysis.
- Demonstrated ability to reduce hallucinations using techniques beyond creative prompting.
- Proficiency in Python and experience with modern ML/AI frameworks.
Benefits
Significant career development opportunities exist as the company grows. The position offers a unique opportunity to be part of a small, fast-growing, challenging and entrepreneurial environment, with a high degree of individual responsibility.
Tiger Analytics provides equal employment opportunities to applicants and employees without regard to race, color, religion, age, sex, sexual orientation, gender identity/expression, pregnancy, national origin, ancestry, marital status, protected veteran status, disability status, or any other basis as protected by federal, state, or local law.
Originally posted on Himalayas