New opportunity

– SDLC GenAI Automation & Tooling Integrations Engineer

M&T Bank · Buffalo, United States of America

About this role

The SDLC GenAI Automation & Tooling Integrations Engineer will modernize M&T Bank’s enterprise software development lifecycle (SDLC) by building GenAI-enabled tooling and integrations. Working with SDLC governance, GenAI engineering, tool owners and delivery teams, the engineer will reduce manual work while improving artifact quality, traceability and control.

Responsibilities

Design, build, test and maintain AI-driven workflows for requirements decomposition, design validation, test planning, artifact creation and validation, approval routing, evidence generation, release notes and governance reporting. Develop agent workflows that identify missing context or poor-quality outputs and route consequential actions for human review and approval.
Create reusable patterns for grounding AI outputs in approved standards, prompt templates, model routing, evaluation, confidence scoring, validation and quality gates. Maintain architectural, security and operational oversight when AI models execute work.
Integrate SDLC platforms through APIs, connectors, scripts, pipelines and event-driven workflows. Connect requirements and Jira work items with GitLab repositories, Zephyr test evidence, ServiceNow permits and RFCs, and reporting systems to support publishing, status synchronization and traceability.
Build data pipelines, telemetry and dashboards for SDLC adoption, artifact completion, approval times, test coverage, defect leakage, control adherence and process efficiency. Monitor automation and integration health, troubleshoot defects, document solutions and support pilots and continuous improvement.

Required qualifications

A bachelor’s degree and at least 5 years of systems analysis or application development experience; an associate’s degree and at least 7 years of that experience; or, without a degree, at least 9 combined years of education and/or relevant work experience, including at least 5 years in systems analysis or application development.
Experience with software architecture, secure development, testing, CI/CD, enterprise integrations, SDLC and Agile/DevOps practices. Hands-on GenAI-assisted development and multi-step AI workflows, including model orchestration, validation, monitoring, exception handling and AI-assisted debugging. Strong communication, documentation and stakeholder collaboration skills.

Preferred qualifications

Experience with SDLC tools such as GitLab Duo, Jira, Zephyr, ServiceNow, SonarQube and Power BI; governance dashboards, agent orchestration and AI observability; or Azure, Kubernetes, Terraform and related cloud infrastructure. Relevant cloud, architecture, AI engineering, DevOps or ITIL training or certifications.

This is a hybrid position in Buffalo, New York. The annual pay range is $116,400–$194,000 USD; benefits described include medical and retirement plans and 40 hours of paid volunteer time each year.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Generative AIAI agentsPrompt engineeringRetrieval-augmented generationContext engineeringModel evaluationModel routingHuman-in-the-loop workflowsAI-assisted debuggingSoftware architectureSystem designAPI designEnterprise integrationsWebhooksEvent-driven workflowsData pipelinesData modelingTelemetryDashboard developmentSecure codingSoftware testingCI/CDAgileDevOpsGitLab DuoGitLabJiraZephyrServiceNowConfluence

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Automate SDLC artifacts and workflows
The role builds and maintains GenAI workflows for requirements decomposition, design validation, test planning, evidence generation, approval routing and governance reporting. Focus preparation on how a multi-step workflow moves an artifact from generation through validation and review.
Preserve quality and human control
The posting calls for grounding outputs in approved standards, evaluating models, detecting missing context and poor-quality artifacts, and retaining human approval and audit evidence. Be ready to explain where quality gates and fail-safe handoffs belong.
Connect systems of record
A central duty is integrating requirements and Jira work items with GitLab, Zephyr evidence, ServiceNow permits or RFCs, and reporting tools. Prepare to describe how APIs, events or pipelines can maintain status synchronization and end-to-end traceability.
Make adoption and control effectiveness measurable
The engineer is expected to lead data pipelines, telemetry and dashboards covering artifact completion, approval cycle times, test coverage, control adherence and automation health. Focus on how metric definitions and source data support useful reporting.
Operationalize and support the tooling
The role includes rapid prototyping alongside secure architecture, monitoring, troubleshooting, runbooks and collaboration with governance, security, tool owners and delivery teams. Prepare to explain how a pilot becomes a maintainable, supportable integration.

Likely challenges to think through

Speed versus consequential review
Likely preparation hypothesis: faster AI-driven publishing or routing could conflict with the required quality, security and approval controls. Prepare a workflow example that distinguishes reversible drafting from consequential actions, with validation, escalation and retained evidence.
Traceability across changing tools
Likely preparation hypothesis: keeping linked artifacts and statuses accurate across Jira, GitLab, Zephyr and ServiceNow may be difficult when updates fail or arrive out of order. Prepare to discuss identifiers, reconciliation, retry or exception handling, and how to avoid silently breaking traceability.
Useful metrics versus noisy telemetry
Likely preparation hypothesis: collecting many dashboard measures may be easier than making them comparable and actionable. Prepare to explain definitions, data provenance and validation for measures such as approval time or evidence quality, alongside operational signals for integration failures.

Practice your answers

Technical practice

How would you design an AI workflow that decomposes requirements into acceptance criteria and test-planning artifacts?

What to coverSuggested preparation: Explain how you would ground outputs in approved standards and templates, identify missing context, and ask for clarification rather than inventing requirements. Describe validation and traceability checks before an artifact is published.

How would you evaluate the quality of generated SDLC artifacts?

What to coverSuggested preparation: Discuss an evaluation approach for completeness, consistency with source requirements, and adherence to quality criteria. Explain how you would use confidence signals and repeatable quality gates without treating a model's confidence as proof of correctness.

How would you orchestrate multiple models and tools in a multi-step agent workflow?

What to coverSuggested preparation: Explain how you would assign generation, validation, and refinement steps based on output quality, speed, and reliability. Address what happens when a step fails, produces conflicting results, or lacks sufficient context.

Where would you require human review in an AI-assisted approval or publishing workflow?

What to coverSuggested preparation: Distinguish low-consequence draft generation from actions that affect approvals, compliance evidence, or downstream systems. Explain fail-safe behavior, review routing, and how you would preserve an auditable record of decisions.

How would you maintain traceability across Jira work items, GitLab changes, Zephyr test evidence, and ServiceNow permits or RFCs?

What to coverSuggested preparation: Describe identifiers, relationships, and synchronization rules that connect artifacts without obscuring their systems of record. Address how you would detect missing links or inconsistent status before producing a traceability view.

How would you build a reliable event-driven integration between SDLC tools?

What to coverSuggested preparation: Compare API, webhook, and pipeline approaches for publishing artifacts or synchronizing status. Cover authentication, duplicate events, retries, failure handling, and tests for data transformations.

How would you design data pipelines and dashboards for SDLC governance and adoption?

What to coverSuggested preparation: Explain how you would define and collect measures such as artifact completion, approval cycle time, test coverage, defect leakage, and control adherence. Discuss data quality and how to distinguish process measures from operational health measures.

What telemetry would you use to operate and troubleshoot an AI-driven SDLC workflow?

What to coverSuggested preparation: Cover workflow throughput, latency, failures, integration health, and output-quality signals, with enough context to trace a problem across steps. Explain how support dashboards and runbooks would help identify whether a fault lies in a model output, integration, or source data.

How would you investigate an AI-generated artifact or code change that appears correct but causes a defect?

What to coverSuggested preparation: Walk through reproducing the issue, checking source context and assumptions, and testing edge cases or vulnerabilities. Explain how structured prompts or a second model could help investigate while human verification remains decisive.

How would you take a GenAI SDLC prototype into production-ready use?

What to coverSuggested preparation: Discuss architecture review, secure development, automated testing, CI/CD, monitoring, and operational support. Explain the trade-off between rapid iteration and maintaining the quality and controls required for an enterprise workflow.

Behavioural practice

Tell me about a time you aligned tool owners and delivery stakeholders on an integration design.

What to coverSuggested preparation: Use situation, action, and result to show how you gathered differing needs, resolved a design decision, and communicated it. Keep the example relevant to cross-team ownership and maintainability.

Describe a time you raised a quality, security, or governance concern about an automation.

What to coverSuggested preparation: Explain the situation, the evidence you used, the action you took with relevant stakeholders, and the result. Show how you balanced delivery progress with appropriate review and control.

Tell me about a pilot or prototype you improved after user feedback.

What to coverSuggested preparation: Set out the initial problem, how you collected and prioritized feedback, what you changed, and the result. Focus on a concrete iteration rather than claiming the first design was already complete.

Describe a production issue you owned across an integration or automated workflow.

What to coverSuggested preparation: Use situation, action, and result to explain diagnosis, coordination, mitigation, and verification. Include how you documented the cause or improved support afterward.

Tell me about a time you made a complex engineering capability easier for others to adopt.

What to coverSuggested preparation: Describe the audience and obstacle, the documentation, example, demo, or support you provided, and the observed result. Connect your actions to clearer usage or more maintainable handoffs.

Sharpen your knowledge

Topics to prepare

SDLC traceability and governance
The posting emphasizes linking requirements, work items, tests, approvals and audit evidence. Suggested revision exercise: diagram how a requirement moves through acceptance criteria, a Jira item, a GitLab merge request, Zephyr evidence and a ServiceNow permit or RFC; mark the records needed to prove each link.
Grounding and evaluating AI-generated artifacts
The role calls for context engineering using approved standards, templates and quality criteria, plus model evaluation and quality gates. Suggested revision exercise: define a scoring rubric for generated acceptance criteria and show how missing source context would prevent publication.
Multi-model orchestration and human control
Required experience includes coordinating multiple AI models and tools in workflows with validation, monitoring and exception handling. Suggested revision exercise: sketch a generate–validate–refine workflow, explaining model-routing tradeoffs and where low-confidence or consequential actions stop for human review.
Secure enterprise integration design
The posting requires API design, secure coding, integration patterns and operational support, including APIs, webhooks, service accounts and event-driven workflows. Suggested revision exercise: draw an artifact-status synchronization design and identify authentication boundaries, failure handling and audit records.
SDLC measurement and operational telemetry
Responsibilities include data pipelines and dashboards for adoption, approval cycle times, test coverage, control adherence and integration health. Suggested revision exercise: define the source records and calculations for two governance metrics and two automation-health signals, noting data-quality limitations.

Skills to revise

Prototype an artifact workflow
The role builds and tests AI-driven requirements decomposition and artifact validation. Suggested hands-on exercise: use sample requirements to produce acceptance criteria, run explicit completeness checks, and route ambiguous outputs to a simulated reviewer rather than publishing them.
Implement cross-tool record mapping
The posting calls for connectors and data transformations linking Jira, GitLab, Zephyr and ServiceNow records. Suggested hands-on exercise: with sample records, map identifiers and statuses between a Jira item, merge request, test evidence and permit; test duplicate and missing-link cases.
Test and debug AI-assisted automation
AI-assisted debugging, root-cause analysis, edge-case identification, vulnerability analysis and output verification are required. Suggested hands-on exercise: introduce a faulty artifact or integration response, use structured prompts to investigate it, then independently verify the fix with tests.
Build a small metrics pipeline
The role builds data models, telemetry and reporting views; Power BI is listed as preferred, not required. Suggested hands-on exercise: transform sample workflow events into artifact-completion and approval-time measures, then display anomalies in a simple dashboard.
Document and support an integration
The posting calls for integration guides, runbooks, troubleshooting and pilot support. Suggested hands-on exercise: write a brief runbook for a failed status-sync event covering detection, diagnosis, safe retry, escalation and evidence preservation.

Your preparation checklist

Map the role’s requirements to evidence
Prepare a one-page map of examples you can discuss for software architecture, secure development, CI/CD, enterprise integrations, GenAI workflows and stakeholder collaboration. Note which examples are hands-on and which tools are only familiar; the posting distinguishes required experience from preferred platforms.
Build a project walkthrough
Select one relevant project, if available, and prepare a concise account of the problem, architecture, your decisions, testing, rollout and results. Show how it reduced manual SDLC work or improved artifact quality, traceability or control; do not imply the project used M&T’s systems.
Practice an agent-workflow design
Sketch a hypothetical requirements-to-test-plan workflow. Show grounding in approved standards, clarification when context is missing, model routing, output validation, quality gates, exception handling and human approval before consequential publishing.
Draw a cross-tool integration
Prepare a diagram connecting a requirement and Jira item to GitLab work, Zephyr test evidence and a ServiceNow permit or RFC. Label APIs or events, identifiers, status synchronization, access controls, failure recovery and the resulting traceability record.
Define measurement and support outputs
Draft a small data model and dashboard mock-up for artifact completion, approval cycle time, test coverage and automation health. Explain data sources, metric definitions, telemetry, alerting and how you would investigate an integration failure.
Rehearse a verified debugging example
Prepare a case showing how structured prompts or multiple models helped investigate a defect, while tests, edge-case checks and your own review verified the fix. The posting specifically calls for AI-assisted debugging with engineering accountability.
Write questions for the interviewers
Bring questions about which SDLC workflow needs a pilot first, where approved standards and systems of record reside, who approves AI-generated actions, and how governance, tool owners and delivery teams will judge quality and adoption.

Availability last checked: Nov 4, 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in United States of America. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…