• Own LLM features end to end: prompts, retrieval, tool use, and APIs
• Build and tune models with classical ML, deep learning, and neural networks
• Ability to use Hugging Face models and datasets for fine-tuning, inference, and evaluation
• Deploy and run models on AWS (including Amazon Bedrock) and Azure
• Set up MLOps for training, versioning, deployment, monitoring, and rollback
• Build RAG pipelines: chunking, embeddings, retrieval, and grounding
• Measure quality with test sets, metrics, and regression checks before release
• Debug latency, cost, hallucination, and data issues in production
• Write clean, tested Python services and integrate them with existing backends