SPEED & SPIRIT is what we look for in our candidates, defined by some simple values that inspire us to BE DRIVEN in our performance, BE VIBRANT in our sporting legacy, BE TOGETHER in our team spirit, and to let our individual talent and experience shine. Applying for a job at PUMA is easy.
Simply click APPLY ONLINE and follow the steps to upload your application.
YOUR MISSIONAbout the Organisation & Role:
As part of the DTC Platform Engineering organisation, you will help deliver the shared platforms, engineering services, and foundational infrastructure that enable product teams to build, test, deploy, and operate software securely and reliably at scale. Joining our AI-Native Engineering Enablement pillar as an AI Engineer, you will focus on production-grade solutions to build, deploy, and scale LLM-powered applications and agentic workflows. In this role, you will bridge the gap between core software engineering, modern machine learning, and tangible business impact—moving generative AI capabilities out of experimental prototypes and into resilient, high-performance production systems for our primary customers: the product engineering teams.
WHAT YOU WILL DO:
Implementing Agentic Workflows
- Construct multi-step AI agents and orchestration workflows using frameworks like LangChain, LlamaIndex, or LangGraph.
- Well-versed with Agentic design patterns.
- Develop production features using foundation models via cloud AI infrastructure
- Deploy and configure enterprise AI and agentic gateways to handle model routing, semantic caching, load balancing, rate limiting, and cost/token tracking.
- Designs and secures the universal integration layer that connects AI models to enterprise databases, internal codebases, and APIs using the Model Context Protocol.
Implement RAG & Retrieval Pipelines
- Build semantic search, document ingestion, and Retrieval-Augmented Generation (RAG) pipelines using vector databases and embedding models,graph databases etc
- Implementation of Vectorless approach wherever required.
Implement AI Memory Architectures:
- Design and maintain advanced cognitive memory systems, including episodic memory, semantic memory, and conversational state management across sessions using graph and in-memory databases.
Production & MLOps:
- Deploy, version, and monitor AI services using containerization and cloud infrastructure.
- Configuring the right serving architecture based on business needs, such as real-time APIs (REST/gRPC), batch prediction jobs, or edge deployment.
- Administering central hubs (like MLflow or Weights & Biases) to track model versions, lineage, code, and artifacts.
- Managing cloud infrastructure costs by auto-scaling compute resources and selecting the right hardware (CPUs vs. GPUs) for production workloads.
- Securing model endpoints, managing access controls (IAM), and ensuring data privacy compliance
Develop AI Test & Evaluation Harnesses:
- Build Automated Evaluation (Auto-Eval) platforms and continuous integration (CI/CD) testing harnesses to stress-test prompts, RAG pipeline outputs, and multi-agent systems at scale.
Cross-Functional Collaboration:
- Partner closely with product managers, engineering manager,SREs ,Infosec team and frontend/backend software engineers, to deliver end-to-end user experiences.
- Bachelor’s or Master’s degree in Computer Science, Software Engineering, Artificial Intelligence, or a related field.
- 4+ years of software development experience, with a dedicated focus on building, testing, and shipping agentic workflows, memory systems, and LLM features to production for 2+ years
- Proven track record of deploying stateful AI systems, complex tool-calling agents, or enterprise API gateway integrations.
- Hands-on experience with multi-agent design patterns, state machines (LangGraph), and stateful memory management tools.
- Practical knowledge of configuring AI gateways for routing, fallbacks, and cost tracking.
- Experience building evaluation harnesses using frameworks like DeepEval, Ragas etc.
- High proficiency in Python (and/or TypeScript for node-based agent runtimes). Rust /Java understanding is good to have.
- Experience with vector stores (Pinecone, pgvector, Milvus, Qdrant) and embedding models.
- RESTful/GraphQL APIs, Docker, Kubernetes, and enterprise cloud infrastructure (AWS, GCP, or Azure).
PUMA provides equal opportunities for all job applicants, regardless of race, color, religion, national origin, sex,
gender identity or expression, sexual orientation, age, or disability. Equality for all is one of the core principles at
PUMA and we do not tolerate any form of harassment or discrimination.
At PUMA, every application is reviewed by real people who are committed to fairness, transparency, and equal
opportunity - no matter your background, identity, or experience.To ensure our process stays true to these values, no automated systems or AI tools are used to make hiring decisions. Every decision is made by real people -with real judgment and accountability. We may use functions supported by Artificial Intelligence (AI) to carry out isolated organizational steps, such as scheduling interviews. These functions have no influence on decisions in the application process. We believe in creating spaces where everyone is welcome, celebrated, and empowered to contribute authentically. Because at PUMA, whoever wants to play, can play.
PUMA is a global sports brand creating footwear, apparel, and accessories that inspire athletes and everyday movers. The PUMA Group owns PUMA, Cobra Golf, and stichd, operates in 120+ countries, and has around 22,000
employees worldwide.
.png){kind=logo}
