Melesh Kesha

• Sole engineer responsible for designing, bootstrapping, and operating three production-grade, air-gapped Kubernetes clusters (Dev/Test/Prod) supporting telecoms infrastructure security assessment at UKTL. No handover from a prior team, as one of the lab's initial employees.
• Production cluster was the verified image source of truth for 20+ active security researchers (penetration testers and vulnerability researchers) assessing telecoms infrastructure, scanning 150 container repositories per week at steady state.
• Built an asynchronous, parallelised automation plane using StackStorm workflows to drive end-to-end JFrog artifact scanning and report generation; pipeline ran without human intervention from trigger to output. Individual Systems Under Test (SUTs) contained up to 2,500 images; parallelised processing reduced full-SUT scan time from 6 hours to 2 hours.
• Engineered decoupled StackStorm trigger actions to automatically synchronise verified SUT container images across isolated network boundaries upon scan completion and policy gate pass.
• Automated zero-touch infrastructure maintenance: weekly JFrog database and vulnerability definition updates were fetched from a single secure gateway and replicated across all clusters.
• Deployed and operated Harbor as the localised container registry; images were scanned and validated before provisioning to the security assessment teams.
• Deployed and operated the platform's stateful backing services: PostgreSQL databases and on-prem S3-compatible object storage underpinning Harbor and JFrog Artifactory within the air-gapped clusters.
• Designed a CI/CT security pipeline integrating JFrog Xray and Tenable Nessus via event-driven orchestration, with a tool-agnostic data store to decouple findings from any single aggregation tool.
• Engineered full-stack observability across all clusters: Prometheus (metrics), Grafana (visualisation), Loki (log aggregation), and Grafana Alloy (centralised telemetry collection and secure log forwarding).
• Designed and executed cluster provisioning via Ansible IaC; declarative state management via ArgoCD.

Key Achievement: Designed, deployed, and operated a complete multi-cluster air-gapped platform scanning 150 repositories per week — including SUTs of 2,500 images processed in 2 hours via parallelised orchestration — serving 20+ telecoms security researchers, built and run as a single-person function from bare hardware to production.

• Developed safety-critical embedded C for Electrically Powered Hydraulic Steering (EPHS) and Electronic Power Steering (EPS) units under ISO 26262 constraints.
• Conducted code reviews, software debugging, and architecture analysis; produced full engineering documentation.
• Built and automated integration test suites using Vector tool suites and CAPL scripting.
• Mentored a graduate engineer on safety-critical compliance frameworks.

Key Achievement: Architected and built frameworks to fully automate existing manual software integration tests, significantly increasing CI velocity.

• Engineered embedded C for automotive instrumentation across CAN and LIN networks.
• Generated automated functional test reports from customer software specifications using Vector VTESTstudio and CAPL scripting.
• Designed a C# application to calculate stack usage safety margins across multiple resource-limited projects dynamically.
• Spent one year embedded in Systems Engineering: requirements elicitation, specification reviews, and verification strategy design.
• Introduced, installed, and configured the software department's SVN version control system and engineering wiki.

Key Achievement: Led an integration testing sub-team focused on knowledge transfer to offshore resources; reduced verification cycle time from 8 hours to 45 minutes.

• Developed unit tests and resolved defects in an internally maintained 3G UMTS protocol stack.
• Practised strict TDD within a globally distributed team; used Perforce and MKS for delivery integrity.

• Sole developer of a C# desktop application to extract, process, and visualise telemetry from the McLaren data analysis package for gearbox reliability evaluation.
• Modified embedded legacy C for sensor calibrations and telemetry, interfacing a legacy ECU with a McLaren SECU via CAN; modified Simulink/StateFlow ECU models for clutch and differential stress testing.

Key Achievement: Self-taught C# to deliver a high-speed telemetry analysis tool under race-team time constraints.