Lead Software Engineer, Full Stack

Capital One Capital One · Banking · New York, NY

Lead Software Engineer focused on building AI-native software engineering harnesses and agentic workflow orchestration systems for governed AI model and coding agent delivery within an enterprise environment. The role involves defining contracts, developing trace capture and observability, integrating evaluation results into CI/CD, and building reusable APIs and tooling to enable AI-native engineering practices.

What you'd actually do

  1. Lead the design and implementation of AI-native software engineering harnesses that uses approved models, coding agents, and developer workflows in governed control-plane capabilities
  2. Build agentic workflow orchestration systems that support planning, code generation, validation, retry loops, human checkpoints, and controlled promotion through delivery environments
  3. Define and implement typed input/output contracts, schemas, metadata capture, and workflow state models that make agentic execution auditable and repeatable
  4. Develop trace capture, observability, and debugging capabilities for AI-assisted engineering workflows, including prompt/output lineage, tool-call traces, model/runtime metadata, and failure analysis
  5. Integrate evaluation results into CI/CD and release workflows, including quality gates for fidelity, build correctness, accessibility, security, performance, and human-review outcomes

Skills

Required

  • software engineering
  • cloud computing (AWS, Microsoft Azure, Google Cloud)
  • JavaScript, Java, TypeScript, SQL, Python, or Go
  • CI/CD, automated testing, quality gates, developer productivity platforms, or observability frameworks
  • Agile delivery environment

Nice to have

  • Master's Degree
  • open-source frameworks
  • cloud-native architectures, microservices, APIs, containers, Kubernetes, serverless platforms, or event-driven systems
  • integrating LLM APIs, building application workflows around generative models, or deploying AI-assisted developer tools
  • building or operating software evaluation frameworks, automated testing runners, regression test suites, or telemetry pipelines, schema-driven development, or workflow orchestration
  • leveraging interactive AI tooling

What the JD emphasized

  • governed AI-native software delivery
  • approved AI models and coding agents
  • enterprise-grade quality, traceability, and operational discipline
  • agentic workflow orchestration systems
  • evaluation results into CI/CD and release workflows
  • governed agentic workflows

Other signals

  • AI-native software delivery
  • agentic workflow orchestration
  • governed AI models and coding agents