Senior AI Engineer - LLM & Agentic Systems

About the role

The Capabilities and Data team within Mastercard Services builds the foundational infrastructure behind customer-facing products, including proprietary large language model ensembles. This role develops and scales shared AI capabilities across Mastercard, improving the scalability, performance and cost efficiency of an existing LLM-based entity resolution system that is already in production. It suits a hands-on AI engineer who combines strong software engineering skills with experience bringing AI solutions into production.

Key responsibilities

• Design and own the architecture for deploying AI capabilities at scale, including APIs and MCP servers for integration with products, agents and applications.
• Build high-performance Python backend services for LLM inference orchestration, async jobs, streaming responses and distributed data workflows.
• Set up LLMOps and MLOps pipelines on Databricks (AWS) with MLflow, feature stores, prompt evaluation, model lineage and continuous retraining.
• Apply traditional ML and statistical modeling alongside LLMs in hybrid architectures.
• Translate R&D outcomes into production solutions, collaborating with researchers to optimize latency, throughput and cost.
• Research and productionize advances in generative modeling, agentic AI, multimodal transformers and foundation models.
• Lead solutions through security and architecture reviews, production approvals and deployment, and work with platform, product and consumer teams.

Required profile

• Masters or Bachelor's degree in Computer Science, AI/ML or Engineering, with significant hands-on experience delivering complex Gen AI or ML programs in production.
• Expert experience designing, building and deploying LLM applications, agentic systems and RAG pipelines from prototype to production.
• Experience building MCP servers or equivalent tool interfaces for LLM and agent consumption.
• Ability to work independently on open-ended problems and communicate cost and architecture trade-offs to engineers and non-technical stakeholders.
• Willingness to follow enterprise production procedures such as security reviews, change management and controlled pipeline promotion.

Required skills

• Deep knowledge of LLM ecosystems (OpenAI, Anthropic, Gemini, Hugging Face, LLaMA, Mistral, Falcon) and Gen AI patterns (prompt engineering, tool/function calling, vector embeddings, semantic search, agent memory).
• Strong Python, including async programming and microservices; SQL required.
• Cloud AI infrastructure (Databricks, AWS SageMaker, Bedrock, Azure OpenAI, GCP Vertex AI) and MLOps/LLMOps tooling (MLflow, Weights & Biases).
• Git, Jira, Visual Studio Code, GitHub Copilot.