I design and run high-performance backends in Go, build cloud-native platforms,
and deploy bare-metal GPU clusters (NVIDIA HGX B200) for LLM training & inference.

• Frameworks & libs: Gin, Echo, Fiber, Chi, net/http (std), gRPC + Protobuf, Connect-RPC
• Data: pgx, sqlc, GORM, Ent, go-redis, MongoDB driver
• Messaging & streaming: Kafka (franz-go, sarama), NATS / JetStream, RabbitMQ
• Architecture: Clean / Hexagonal architecture, DDD, CQRS, Event Sourcing, Saga, Outbox pattern
• Concurrency: goroutines & channels, worker pools, context propagation, rate limiting, backpressure
• Performance: pprof, trace, benchmarks, escape analysis, memory/GC tuning
• Reliability: circuit breakers, retries with jitter, idempotency, graceful shutdown
• Testing: testify, gomock, testcontainers-go, fuzzing, table-driven tests
• Tooling: Go modules & workspaces, golangci-lint, Wire / Fx (DI), Cobra CLIs, Kubernetes operators (controller-runtime, Kubebuilder)

• Containers & orchestration: Docker, Kubernetes (kubeadm, k3s, EKS/GKE), Helm, Kustomize
• GitOps & CI/CD: Argo CD, Flux, GitHub Actions, GitLab CI, Argo Workflows / Rollouts (canary, blue-green)
• Infrastructure as Code: Terraform / OpenTofu, Ansible, Packer, cloud-init
• Networking: Cilium (eBPF), Calico, Istio / Linkerd, Ingress-NGINX, Traefik, MetalLB, HAProxy, Envoy
• Observability: Prometheus, Grafana, Loki, Tempo, OpenTelemetry, Alertmanager, VictoriaMetrics
• Storage: Ceph / Rook, Longhorn, MinIO (S3), NFS, local NVMe
• Security: Vault, cert-manager, external-secrets, Trivy, Kyverno / OPA, RBAC, network policies, SSO (Keycloak)
• SRE practices: SLO/SLI, incident response, postmortems, capacity planning, chaos testing
• Linux: kernel & sysctl tuning, systemd, networking (iproute2, bonding, VLANs), bash automation

End-to-end bring-up of bare-metal GPU servers: from rack to running LLM endpoints.

Hardware & bring-up

• NVIDIA HGX B200 (8× Blackwell GPU, 192 GB HBM3e each, 5th-gen NVLink / NVSwitch)
• Rack planning: power (high-density PDUs, ~14 kW+ per node), cooling (air / direct liquid cooling), cabling
• BMC / IPMI / Redfish, BIOS & firmware updates, PXE / MAAS bare-metal provisioning
• Burn-in & validation: DCGM diagnostics, nvidia-smi, NCCL tests, HPL, GPU stress tests

GPU networking

• InfiniBand NDR 400G & RoCEv2 (Spectrum-X), NVIDIA ConnectX-7 / BlueField-3 DPU
• Rail-optimized fat-tree topology, UFM, Subnet Manager, GPUDirect RDMA & GPUDirect Storage
• NCCL tuning for multi-node all-reduce performance

Software stack

• NVIDIA drivers, CUDA 12.x, Fabric Manager, NVIDIA Container Toolkit
• Kubernetes GPU Operator, Network Operator, MIG partitioning, time-slicing
• Scheduling: Slurm (Pyxis / Enroot), Kueue, Volcano, Run:ai concepts
• Monitoring: DCGM Exporter → Prometheus / Grafana (temps, power, ECC, XID errors, NVLink health)
• High-performance storage: WEKA / VAST / Lustre concepts, NVMe-oF

LLM serving & training

• Inference: vLLM, SGLang, TensorRT-LLM, NVIDIA Triton, NVIDIA NIM, llama.cpp, Ollama
• Optimizations: FP8 / FP4 quantization, tensor & pipeline parallelism, continuous batching, KV-cache tuning, speculative decoding
• Distributed training: PyTorch DDP / FSDP, DeepSpeed, Megatron-LM, LoRA / QLoRA fine-tuning
• MLOps: Ray, KServe, Kubeflow, MLflow, LiteLLM gateway (OpenAI-compatible APIs)

• LLM APIs: Anthropic Claude, OpenAI, Gemini, open-weight models (Llama, Qwen, DeepSeek, Mistral)
• RAG pipelines, embeddings, hybrid search, reranking
• Vector DBs: pgvector, Qdrant, Milvus, Weaviate
• AI Agents, tool calling, MCP (Model Context Protocol) servers in Go & TypeScript
• LangChain, LangGraph, LlamaIndex, Vercel AI SDK
• Evaluation & observability: Langfuse, LangSmith, OpenTelemetry for LLMs

🎨 Frontend

• TypeScript, React 19, Next.js 15 (App Router, RSC), Vue 3, Nuxt 3
• TanStack Query / Router, Redux Toolkit, Zustand, GraphQL (Apollo)
• TailwindCSS v4, shadcn/ui, Radix, MUI, Ant Design
• Testing: Vitest, Jest, React Testing Library, Playwright, Cypress

🧩 Other Backend

• Node.js, NestJS, Express, Hono, Bun
• PHP 8.3+, Laravel 12, Yii2
• MySQL, Oracle, PostgreSQL, MongoDB
• Prisma, Drizzle ORM

📱 Mobile

• React Native, Expo
• Swift, SwiftUI
• Ionic