Ayush Baranwal

I am a Software Engineer at Deutsche Bank with around 3 years of experience in backend engineering, distributed systems, CI/CD platforms, and cloud-native infrastructure.

My work focuses on building scalable and reliable systems, including GitHub Actions infrastructure, observability platforms, backend services, automation, and compliance-driven workflows. I have hands-on experience with Java, Python, Spring Boot, FastAPI, REST APIs, Kubernetes, Docker, GCP, Prometheus, Grafana, Jenkins, TeamCity, Oracle, and SQL.

At Deutsche Bank, I have worked on enterprise-scale systems supporting 1,000+ GitHub organizations, 8–10K repositories, and millions of workflow job runs per month. I enjoy solving complex engineering problems around system design, reliability, performance, and developer productivity.

I am currently exploring Software Engineer / SDE-2 opportunities in backend engineering, platform engineering, distributed systems, and SRE-focused roles.

Software Engineer at Deutsche Bank (2023-07 – Present)

• Architected and implemented an enterprise-scale GitHub Actions observability platform supporting 1,000+ organizations, 8–10K repositories, and 4M workflow job runs/month during ARCv1 → ARCv2 migration.
• Designed ingestion architecture after evaluating webhooks vs polling, GitHub PATs vs GitHub Apps, and REST vs GraphQL APIs, enabling scalable data collection without workflow-level changes.
• Built a Python/FastAPI aggregation service generating Prometheus metrics and delivered 3 production-grade Grafana dashboards (Migration Overview, Ops & Efficiency, Org/Repo Drilldown) for adoption tracking, queue/run-time analysis, and runner utilization insights.
• Architected and deployed org-level ARCv2 self-hosted runner infrastructure across DEV/UAT/PROD using Helm-driven multi-stage workflows (controller charts, external-secrets sync with Google Secret Manager, runner scale-sets), provisioning 10 dedicated runners and reducing CI queue time by 70% via workload isolation from the shared enterprise runner pool.
• Migrated GitHub Enterprise audit logs to a standardized GCP landing zone (GCS, IAM, KMS) and integrated with downstream security/compliance pipelines to meet regulatory and audit requirements.
• Contributed to SDLC compliance platform (Factstore) implementing policy-driven rule evaluation (5 policies/release, 4–5 rules each) with PR-level enforcement using a last compliant commit hash mechanism.
• Designed and delivered Access Management (dual-role, active/inactive), centralized audit logging, and Cucumber-based test automation, reducing test setup time from 1 hour to 4–5 minutes for a 7-member engineering team.
• Developed a scalable tool registration framework abstracting CI/CD integrations (GitHub/Bitbucket, Jenkins/TeamCity), enabling onboarding of 100 internal applications.

Research Intern at Queen Mary University of London (2023-05 – 2023-07)

• Designed and developed EdgeAISim, a Python-based simulation toolkit leveraging Reinforcement Learning (RL) and Graph Neural Networks (GNNs) for energy-efficient task scheduling in edge computing environments.
• Implemented dynamic task migration and workload optimization strategies, achieving 58% reduction in power consumption compared to baseline scheduling methods.
• Co-authored and published findings in Elsevier (Measurement: Sensors, 2024), contributing to research on AI-driven energy optimization in distributed systems.

B.Tech in Information Technology – IIIT Allahabad (2019-07 – 2023-05)