Dmitrii
Parshenkov

Ketryx Corporation

Maintained and evolved the AWS infrastructure of a regulated medtech SaaS scaleup across 30+ customer-facing environments. Drove major platform migrations, infrastructure cost reduction, and the company's security and compliance operations.

• Led the platform's migration from Elastic Beanstalk to ECS to completion across 31 single-tenant enterprise and multi-tenant production environments: containerized the monolith, centralized environment states into a single deployment repository, and implemented job-queue and request-based autoscaling.
• Cut monthly infrastructure spend by ~60%, from ~$260k to ~$106k, with no reduction in workloads or reliability: fleet-wide rightsizing of compute, storage, and autoscaling, scheduled weekend wind-downs, and an ~80% cut in CI spend ($60k+ down to ~$12k/month).
• Rebuilt GitHub Actions CI/CD on self-hosted RunsOn runners (replacing Depot), orchestrating 30,000+ EC2 instances daily: spot placement strategies, per-workload instance sizing, and aggressive caching (ECR pull-through, Docker layers, GHA cache).
• Optimized the CI/CD pipelines with duration-weighted E2E test bin-packing, multi-level caching, parallel execution, and reusable workflows, cutting the release pipeline from ~59 to ~42 minutes on half the VMs and ~20% less compute.
• Drove the ISO 27001:2022 and SOC 2 audits end-to-end: prepared and defended control evidence with external auditors, maintained ISMS documentation, remediated nonconformities via CAPAs, and commissioned an external penetration test. Owned ongoing compliance operations: automated control monitoring, vendor reviews, personnel HIPAA/GDPR training, and customer security questionnaires.
• Unified infrastructure management under Terragrunt across all single and multi-tenant environments: imported all ClickOps and CloudFormation-managed resources (RDS, ElastiCache, VPCs, CloudFront, WAF, SES, security tooling), enforced cost-allocation tags, and defined SSO permission sets, KMS/SOPS secrets, and GitHub repo config.

MOSTLY AI Solutions MP GmbH

Dual role across Platform Engineering and Information Security in a high-velocity startup. Operated and evolved the Kubernetes platform for AI/ML workloads. Maintained ISMS policies, enforced controls, and resolved non-conformities.

• Completely revamped the existing Kubernetes platform, introducing a multi-environment GitOps-driven solution (ArgoCD, later FluxCD) with enhanced RBAC (EKS SSO, IRSA) and networking (AWS LBC, Readiness Gates) for efficiency and security.
• Successfully consulted through an ISO 27001:2022 review and prepared for the SOC2 Audit.
• Implemented sophisticated Kubernetes auto-scaling infrastructure on EKS via Karpenter, Pod Preemption and Hot Pools to support AI Agents/ML Jobs, raising steady-state cluster memory utilization to ~75-85% with headroom reserved for overcommitment.
• Ensured platform availability via a modern Operator-based monitoring stack: Prometheus/Alertmanager (via Robusta), OpenTelemetry (Tempo), and Loki (via Alloy), all managed via IaC.
• Built a fully-automatic GitOps-based preview environments system, cutting environment provisioning from days (long-lived branch deploys) to minutes per pull request while significantly lowering infrastructure costs.
• Reworked cloud infrastructure and networking by migrating to IaC via Terragrunt over OpenTofu (and CrossPlane), enhancing maintenance and observability.
• Refactored GitHub Actions CI/CD with remote workflows and custom internal builders (via ARC), improving integration performance and security.
• Introduced Helm-charts for internal/customer use with an in-house library and schema validation, redefining enterprise installation procedures.

MightyCall (CallCurrent Inc.)

Designed and built the company's new infrastructure platform, migrating from the Hashicorp stack to Kubernetes. Managed a team of 3 engineers, a hybrid fleet of over 200 servers and owned the cloud budget. Responsible for the reliability of the production environment and the efficiency of the developer experience.

• Established multi-environment CI/CD (via GitOps on ArgoCD) and access control, enabling a seamless transition from the legacy Orchestration Stack.
• Introduced SaltStack to replace Ansible for autonomous dependency and update management across Linux and Windows machines, ensuring consistency.
• Led the implementation of a modern observability stack: metrics via Prometheus, tracing via OpenTelemetry, and centralized logging via Loki, unified in Grafana.
• Adopted Skaffold for local development, featuring pre-configured components to allow for rapid iteration and clean testing, drastically speeding up the SDLC.
• Redesigned configuration systems for legacy components, introducing DRY principles with parametrization to improve maintainability.

MightyCall (CallCurrent Inc.)

Operated the Hashicorp stack (Nomad, Consul, Vault) and built IaC-driven environments via Terraform and Ansible, enabling self-service deployments for dev and QA teams.

• Designed an in-house API Gateway and Load Balancing Solution based on Envoy-proxy, utilizing Consul KV and Consul Template for its dynamic configuration.
• Performed a complete overhaul of the CI/CD Jenkins pipelines and transitioned to the Declarative Pipeline approach - introduced Groovy Libraries (Classes, Functions) for standardization and easier maintenance/quality control. This also sped up deployment/build times dramatically (1 hour to 8 minutes on average per component), whilst at the same time heavily improving compute usage.
• Introduced monitoring setup as part of the deployment process, configuring Zabbix integration (via Zabbix Agent) for base health metrics and simple alerts and New Relic for APM (Application Performance Monitoring).

MightyCall (CallCurrent Inc.)

Held alongside the DevOps Engineering role. Managed a team of 8 remote support agents and led the Anti-Fraud department. Improved internal and cross-departmental communication, positioning Support as a customer-centric insights provider for PM and Marketing while facilitating QA/Development's engagement with Tier 2 on existing bugs.

• Introduced new onboarding and qualification improvement processes - erasing the gap between Tier 1 and Tier 2 teams, improving turnaround times and response quality.
• Completely reworked the working schedule, reducing the missed support call ratio from 86% to 4% while maintaining the same working hours.
• Maintained and improved support tooling (Zendesk, LiveChat, Intercom, etc.), introducing new automations, triggers, new ways to interact with customers and better ways to handle customer feedback, as well as perform analytics.
• Conducted daily routine checks and educated an AI-driven self-learning anti-fraud platform. Collaborated with providing carriers and FBI to maintain a fraud-free service.