A four-day intensive workshop designed for computer science research scholars and faculty. Unlike typical AI training that stops at API usage, this program takes participants through the complete pipeline of building, adapting, and deploying AI systems with runnable code and free tools. On Day 1, participants explore large language models (LLMs) and agentic AI, learning how to design agents that reason, retrieve knowledge, and coordinate under human oversight. Day 2 introduces the Model Context Protocol (MCP), enabling safe and structured integration of external tools, databases, and APIs. Day 3 focuses on fine-tuning with parameter-efficient methods like LoRA/QLoRA, allowing specialization of models on domain-specific data at minimal cost. Finally, Day 4 covers edge AI and compression, teaching quantization and distillation techniques to deploy models offline on laptops, phones, or Raspberry Pi devices.

By the end, participants gain skills in agentic systems, reproducible pipelines, efficient deployment, and cost-independent AI research. The workshop emphasizes practical outcomes such as publishable contributions, and domain-specialized models for fields like healthcare, agriculture, and education. new research directions in low-resource and privacy-preserving AI.