OCI Generative AI Professional (1Z0-1127)

Learning plan

About 40 hours of focused study, split into 5 phases. Pick your weekly pace and the timeline adjusts.

Estimated finish: 4 weeks

Before you start

• Check eligibility first: Oracle archived the 2025 Generative AI Professional learning path in 2026, and its August 2026 update lists OCI Enterprise AI Professional as the professional-level agent certification. Confirm which exam is bookable on Oracle University before you start.
• Basic ML/deep-learning concepts and working Python
• An OCI account in a region where the Generative AI service is available. Dedicated AI clusters for fine-tuning are paid

1. Week 1 · 8 h

   LLM fundamentals

   • Explain encoder, decoder and encoder-decoder transformers, and which tasks suit each
   • Use prompting strategies (zero/few-shot, chain-of-thought, least-to-most) and recognise prompt injection
   • Compare prompting, RAG and fine-tuning by cost, effort and data needs, and explain decoding settings (temperature, top-p, top-k)
   • Define hallucination, groundedness and attribution, and how RAG reduces hallucination

   Use: Hugging Face LLM Course (free)

2. Weeks 1–2 · 11 h

   OCI Generative AI service

   • Use the playground and model catalogue, and tell chat models from embedding models
   • Compare on-demand inference with dedicated AI clusters (hosting vs fine-tuning units) and how billing differs
   • Explain fine-tuning methods (T-Few vs Vanilla), training-data format and the effect of fine-tuning on endpoints
   • Secure the service with IAM policies and compartments, and know that customer data is not used to train base models

   Hands-on: In the OCI Generative AI playground, run one prompt against two chat models, export the generated Python SDK code, and call the model from a local script with your own temperature and max-tokens settings.

   Use: OCI Generative AI documentation

3. Weeks 2–3 · 10 h

   RAG with Oracle AI Vector Search and LangChain

   • Chunk documents, generate embeddings with OCI Generative AI, and store them in a VECTOR column
   • Run similarity search with VECTOR_DISTANCE and approximate indexes (HNSW, IVF), and pick a distance metric
   • Build a LangChain RAG chain that uses Oracle AI Vector Search as the vector store and an OCI chat model, with conversation memory

   Hands-on: On an Always Free Autonomous Database, load a few PDFs, embed the chunks, store them in a VECTOR column, and build a LangChain chatbot that answers questions and cites the source chunks.

   Use: Oracle AI Vector Search guide · LangChain documentation

4. Weeks 3–4 · 5 h

   Agents on OCI

   • Explain OCI's managed agent offering (knowledge bases, agents, endpoints, tools) and how it grounds answers in your data
   • Describe tool/function calling and how Oracle's 2026 Enterprise AI agents use the Responses API, MCP and guardrails

   Use: OCI 2026 certification updates (Oracle University blog) · Oracle Cloud Free Tier

5. Week 4 · 6 h

   Practice and review

   • Take Oracle's practice exam for whichever professional exam is current and review every miss
   • Memorise the dedicated-cluster unit types and fine-tuning method differences, which are frequently tested
   • Re-run your RAG lab from scratch without notes

   Use: Udemy – OCI Generative AI Professional practice tests

You're ready to book when…

• You can explain T-Few vs Vanilla fine-tuning and on-demand vs dedicated serving, with cost implications
• You have built a working RAG chatbot on Oracle AI Vector Search
• You score 75% or higher on a timed practice exam for the current exam code

Exam tips

• Oracle's August 2026 update removed calendar years from certification titles and made OCI certifications valid for 24 months from the date earned.