Mehdi Baneshi

Designed and operated cloud-native platforms on GCP and AWS, from initial architecture through production operations. Full ownership of infrastructure decisions, cost optimization, and reliability.

• •Architected hexagonal / clean architecture systems with zero domain-layer coupling, enabling swappable adapters for databases, APIs, and message queues
• •Designed dual-mode execution platforms running identically on local Docker and GCP Cloud Run, eliminating environment drift
• •Led microservices migration from monolith — decomposed into 12 independent services with event-driven communication via Pub/Sub
• •Built and maintained Kubernetes clusters (GKE) serving production workloads with auto-scaling and zero-downtime deployments
• •Implemented Infrastructure as Code (Terraform) managing 50+ cloud resources across staging and production environments
• •Reduced cloud infrastructure costs by 40% through right-sizing, preemptible instances, and architectural optimizations

Built CI/CD pipelines and automation systems across every project. Treat infrastructure and deployment as first-class engineering problems, not afterthoughts.

• •Established CI/CD pipelines (GitHub Actions, Cloud Build) reducing deployment time from 45 minutes to under 8 minutes with automated testing gates
• •Designed automated deployment scripts and release workflows used across 15+ projects and 3 teams
• •Implemented comprehensive monitoring stacks (Prometheus, Grafana, Cloud Monitoring) reducing incident response time by 70%
• •Built automated data pipelines processing 5GB+ daily ingestion with ETL workflows and alerting
• •Created developer tooling and internal CLIs that reduced onboarding time for new engineers from days to hours
• •Managed Linux infrastructure, networking (VPN, DNS, firewalls), and security hardening across all environments

Building production AI systems — not prototypes. Focus on agent orchestration, RAG pipelines, and integrating LLMs into real workflows with reliability and observability.

• •Built multi-agent orchestration system using Google ADK with self-registering agents processing 23 signal categories autonomously
• •Engineered event-driven AI pipeline: Capture, Classify, Enrich, Decide, Execute — with human-in-the-loop oversight at decision boundaries
• •Integrated LLM-powered features (document summarization, semantic search, automated triage) into production SaaS platform, increasing user engagement by 35%
• •Designed RAG pipelines with vector databases (ChromaDB, Pinecone) for domain-specific knowledge retrieval
• •Deployed local LLM inference (Ollama) alongside cloud APIs (Claude, GPT, Vertex AI) for cost-optimized AI workloads
• •Built prompt engineering frameworks and evaluation harnesses for consistent, measurable AI output quality

End-to-end product development across web, mobile, and desktop. Comfortable as the sole engineer or leading a small team. Ship fast, architect for scale.

• •Delivered 15+ production applications across fintech, healthcare, e-commerce, and developer tools
• •Built real-time trading dashboard processing 10K+ events/second with WebSocket architecture, serving 200+ institutional traders
• •Developed cross-platform mobile apps (React Native, Swift, Kotlin) with 50K+ downloads and 4.7-star ratings
• •Designed and implemented OAuth2/OIDC authentication systems adopted across 8 client projects
• •Built RESTful and GraphQL API layers handling 100K+ daily requests with sub-200ms p95 latency
• •Established shared component libraries and design systems reducing new project bootstrap time by 60%
• •Increased test coverage from 20% to 85% across team projects by introducing TDD practices and automated testing gates