Job Summary
We are seeking technical and experienced AI Agentic Solutions Architects to join our advanced technology group. This role is dedicated to the hands-on design, development, and deployment of sophisticated Agentic AI solutions. You will be responsible for executing complex Proof of Concepts (PoCs) that demonstrate cutting-edge AI capabilities for our key customers. The role requires building and optimizing AI agents on high-performance Nvidia-based GPU infrastructure (including A100 and H100 systems) and leveraging Cognizant's proprietary ATK platform. The ideal candidate is a hands-on architect with a strong background in MLOps and extensive experience with modern agentic frameworks, advanced data semantics, and the Nvidia AI ecosystem.
Key Responsibilities
Architect, build, and optimize multi-agent AI solutions using frameworks like LangChain, LlamaIndex, and AutoGen.
Design and implement complex data ingestion and processing pipelines, ensuring robust data semantics for Retrieval-Augmented Generation (RAG) architectures.
Develop and fine-tune Large Language Models (LLMs) and other foundational models using the Nvidia NeMo framework.
Deploy and manage high-throughput, low-latency model inference services using Nvidia Triton Inference Server.
Conduct performance profiling and optimization of AI workloads on Nvidia A100 and H100 Tensor Core GPUs.
Integrate and manage specialized vector databases such as Milvus, Pinecone, and Weaviate for high-dimensional data indexing and search.
Leverage Cognizant's ATK platform to orchestrate complex agentic workflows and ensure seamless integration with enterprise systems.
Collaborate with infrastructure teams to ensure optimal configuration of Kubernetes clusters for GPU-accelerated workloads.
Required Skills and Experience
Agentic Frameworks: Minimum 2+ years of hands-on experience with LangChain and/or LlamaIndex. Demonstrable ability to build complex chains, tools, and autonomous agents.
Vector Databases: Proven expertise in deploying and managing at least one of the following: Milvus, Pinecone, Weaviate, or ChromaDB. Deep understanding of embedding models, indexing strategies (e.g., HNSW, IVF), and semantic search.
Nvidia Software Stack:
Nvidia Triton Inference Server: Hands-on experience deploying and scaling models for production.
Nvidia NeMo: Experience in training or fine-tuning models.
CUDA/cuDNN: Strong understanding of the CUDA programming model and library ecosystem for GPU acceleration.
Nvidia Hardware Stack: Demonstrable experience working with and optimizing for Nvidia A100 or H100 GPUs. Familiarity with concepts like Multi-Instance GPU (MIG) is highly desirable.
ML Frameworks & MLOps: Expert-level proficiency in Python and PyTorch. Strong experience with MLOps principles and tools, including containerization with Docker and orchestration with Kubernetes.
Problem-Solving: Advanced analytical and debugging skills for complex, distributed AI systems.
Preferred Qualifications
Experience with advanced RAG techniques, including re-ranking and query transformation.
Familiarity with Nvidia's full enterprise software suite.
Experience in a customer-facing role, delivering technically complex PoCs or solutions.
Active contributions to open-source projects in the AI/ML ecosystem.
关于高知特 (Cognizant)
高知特(Cognizant)（纳斯达克代码：CTSH）作为一家AI Builder和相关技术服务提供商，致力于通过打造全栈AI解决方案，帮助企业将人工智能投资转化为实际价值。公司凭借深厚的行业经验、流程优化和工程技术专长，将企业独特的业务场景融入科技系统，赋能组织释放人才潜能，推动切实成果，并帮助全球企业在瞬息万变的环境中保持领先。如需了解更多详情，敬请访问 cognizant.ai 或关注@cognizant。
补充雇佣信息
薪酬信息截至本职位发布之日为准确。Cognizant 保留在适用法律允许的范围内随时修改该信息的权利。
申请人可能需要通过现场面试或视频会议的方式参加面试。此外，候选人在每次面试时可能需要出示其当前所在州或政府签发的有效身份证件。
Cognizant 是一家提供平等就业机会的雇主。在招聘过程中，您的申请和候选资格不会因种族、肤色、性别、宗教、信仰、性取向、性别认同、国籍、残疾、遗传信息、怀孕、退伍军人身份或任何其他受联邦、州或地方法律保护的特征而受到影响。
