Sadiqi Omar

Member of the Data & AI team: led the migration of 29 legacy research clusters to a modern OpenSearch platform on Kubernetes, a search, analytics, and vector store foundation powering the bank's Data & AI use cases, distributed across 3 on-premise datacenters with full security compliance (AICP / CVAC). Program conducted internationally, across Asia, Europe, and the United States.

- AI Platform: OpenSearch utilized as the bank's AI foundation — MCP servers, RAG / LLM orchestration, local knowledge bases, and model vectorization / embeddings (vector store).

- GitOps: Deployment and operation of OpenSearch clusters via GitOps (ArgoCD, Kustomize) on DEV, PREPROD & PROD.

- End-to-end Security: TLS certificates, LDAP / Active Directory authentication, at-rest encryption, and secret management via HashiCorp Vault.

- Blue-Green Strategy: Upgrade by snapshot/restore ensuring zero data loss during major version migrations.

- Observability: Full stack Prometheus, Grafana, Alertmanager with automated incident routing to ServiceNow.

- Handover & Run: Operational documentation, runbooks, and on-call procedures; onboarding of Tech Leads and Architects.

Stack: OpenShift, Kubernetes, ArgoCD, Kustomize, OpenSearch, HashiCorp Vault, Prometheus, Grafana, ServiceNow.

Cloud Tech Lead and Senior Data Engineer at Sodexo DataHub: end-to-end management of industrialization technical initiatives, from Cloud & AI component integration to production deployment, with a strong PySpark / Databricks on Azure culture and a Design-to-Dev approach.

- Industrialization & Production Deployment: Design, development, and optimization of ETL data pipelines with PySpark — systematic transition from POC to production.

- CI/CD & Automation: Deployment of CI/CD pipelines via Azure Pipelines, versioning, and development cycle automation.

- Real-time Processing: Streaming solutions with Azure EventHub and PySpark.

- Observability: Real-time monitoring and alerting with Grafana & Prometheus — foundation reused for monitoring AI service costs and quality.

- Query Optimization: Strategic use of Dremio in parallel with PySpark to accelerate queries.

Site Reliability Engineer within one of the five global Product Reliability Engineering teams: responsible for the reliability, application performance, and large-scale technical support.

- Reliability & Observability: Monitoring of a large application park (Python, Bash, Groovy) with Prometheus, Grafana, Kibana, Graphite, and VictoriaMetrics — expertise directly transferable to LLMOps observability.

- ML Model Observability: Implementation of production Machine Learning product monitoring — detection of drift (data & model drift), tracking of quality metrics, and alerting to ensure reliable predictions over time.

- Incident Management: Active participation in on-call rotations, real-time alert management, and rapid incident resolution to maximize service continuity.

- Orchestration: Kubernetes, Mesos & Marathon for deployment management; Apache Spark for large-scale data processing on Linux.