LLMOps
GenAI Platform Engineering for open-weight LLMs — GPU sizing, model packaging, vLLM/TGI/SGLang serving, RAG with Qdrant, Docker/Kubernetes, CI/CD, monitoring, Ray scale, and continuous improvement. Cloud-agnostic LLMOps lifecycle.
4.9
(290 students)
8 Weeks · 48 hours
Next Cohort:
No upcoming sessions scheduled yet.
Can't find a suitable date?
What is this course about?
LLMOps
GenAI Platform Engineering · Open-Weight Models · Serving · RAG · Kubernetes · CI/CD · Monitoring
Operationalize open-weight large language models end to end—package and size GPUs, serve with vLLM/TGI/SGLang, build retrieval pipelines, containerize and orchestrate on Kubernetes, automate CI/CD evaluation gates, monitor drift, and run continuous improvement with governance.
What you'll walk away with:
- LLMOps lifecycle fluency from checkpoint → packaging → deployment → monitoring → retraining
- GPU sizing and packaging skills across Safetensors, GGUF, GPTQ, AWQ, and ONNX
- Serving benchmarks with vLLM, TGI, SGLang, and llm-d for distributed inference
- RAG/CAG/agentic pipelines on an open-source stack (Qdrant + open embeddings + local models)
- Docker/Kubernetes deployment patterns with GPU scheduling and autoscaling
- CI/CD, observability, A/B experimentation, and governance for production LLM platforms
Course Features
Post Graduate Diploma
8 Weeks of Content
Hands-on Projects
Community Support
Lifetime Access
Student Reviews
Arjun Mehta
Best LLMOps program I’ve taken—GPU sizing, vLLM vs TGI benchmarks, and Kubernetes GPU scheduling finally clicked. Now running open-weight serving at work.
Infosys LimitedNisha Verma
The RAG + Qdrant labs and CI/CD evaluation gates mirror what our platform team needed. Capstone runbook became our internal rollout checklist.
Wipro TechnologiesSiddharth Iyer
Packaging formats (GGUF/AWQ) and speculative-decoding labs were gold. Moved from ML experiments to a production LLM platform role.
TCSPooja Nair
Clear path from lifecycle mapping to Ray Train + drift monitoring. Interviewers loved the A/B promotion and governance labs.
AccentureRahul Deshmukh
Cloud-agnostic stack focus is rare—Docker, K8s, llm-d, and cost-per-token sizing without vendor lock-in. Highly practical.
CognizantAnanya Krishnan
Serving benchmarks and autoscaling under load prepared me for on-call. Promoted to GenAI Platform Engineer within a quarter.
Capgemini IndiaJames Whitfield
Excellent LLMOps depth—evaluation gates, model registry rollback, and multi-armed bandit experiments. Now leading open-weight inference at a US SaaS company.
DatabricksSofia Alvarez
From Hugging Face checkpoint to monitored Kubernetes deployment in one curriculum. The labs are portfolio-ready and interview-proof.
NVIDIA
