Lead Software Engineer - Java or Python, Agentic AI

As a Lead Software Engineer at JPMorgan Chase within the Commercial & Investment Bank Digital Channel's team, you will build autonomous agent capabilities that can plan, execute, validate, and submit code changes in bulk . The role focuses on the evaluation harnesses, PR-provenance controls, CI/CD integrations, and operational readiness needed to scale machine-authored changes safely across runtime upgrades, framework migrations, security remediation, and standards adoption.

Job Responsibilities

• Design and integrate AI-driven remediation workflows into enterprise CI/CD pipelines, including trigger design, build and test gating, deployment readiness checks, failure handling, and evidence capture for downstream audit and operational review.

• Build and operate the evaluation harness that proves agent quality and delivery readiness at scale, including success rate, regression rate, PR-merge rate, pipeline pass rate, drift detection, and control effectiveness across runtime, framework, standards, and CVE-remediation skills.

• Implement the PR-provenance contract end-to-end under the Sr Lead's design, including branch creation, CI/CD hook integration, build-verification gates, automated test evidence, audit-trail emission, signed commit and merge attestation, rollback envelope, and operational handoff for failed or blocked runs.

• Own specific subsystems within the harness, including evaluation-fixture management, replay tooling, regression corpora, quality-signal aggregation, failure triage, runbook maintenance, and day-to-day operational support.

• Co-own the agent harness's reliability and observability including metrics, logs, traces, replay tooling, alerting, dashboards, and failure-pattern analysis so agent behavior and pipeline outcomes are diagnosable at scale.

• Contribute to agent-skill design reviews as an SME-capable second pair of eyes on eval-coverage and provenance implications; escalate audit-control questions to the L5.

• Support the Standards pillar and Tooling pillar by wiring at-scale rollout of new lint rules, template upgrades, quality gates, and migration checks into the evaluation harness and CI/CD flow so bulk agent runs can prove standards adoption at Channels scope.

• Instrument value, adoption, and operational metrics for the evaluation and provenance subsystems, including number of repositories evaluated, number of PRs provenance-signed, pipeline pass and failure rates, repeat-run reduction, engineering days saved, and audit-evidence completeness.

• Drives team adoption of enterprise-authorized AI-assisted engineering practices within the work environment to improve code quality, delivery speed, and operational outcomes (e.g., AI-assisted code review/refactoring, test strategy acceleration, incident/root-cause analysis support), while establishing consistent validation standards (secure coding, peer review, automated testing) and promoting reuse of effective patterns across the team.