watsonx GenAI Engineer case studies

watsonx GenAI Engineer · Put your knowledge to work in context.

The watsonx Generative AI Engineer credential is aimed at practitioners who build, tune, govern, and operate generative AI solutions on the IBM watsonx stack. It suits engineers and architects who already write code, call APIs, and reason about data pipelines, and who now need to demonstrate they can take a foundation model from prompt to production responsibly. The watsonx GenAI Engineer case studies are 5 cases and about 4,071 words.

Invoice extraction at 40,000 calls a day: few-shot, schema, or fine-tune?

A shared-services team extracts 12 fields (vendor, invoice number, dates, currency amounts, PO number, tax, payment terms) from 40,000 invoices per day. The current design is a zero-shot prompt on Granite-13b-instruct with a 1,900-token instruction block, JSON parsed by regex with a retry-on-parse-failure fallback.

826 words, about 4 min 2

Recall halves after an embedding model upgrade in the policy assistant

An enterprise knowledge assistant answers HR and finance policy questions over 85,000 PDFs, chunked into 1.2 million passages of 512 tokens with no overlap and stored in a Milvus collection in watsonx.data. The team recently replaced a 768-dimension embedding model with a stronger 1,024-dimension model and rebuilt the index, but did not change chunking.

799 words, about 4 min 3

A 14-category clause classifier with noisy labels and three weeks

A legal operations team wants to classify contract clauses into 14 categories and extract 5 attributes (party, effective date, term length, termination notice, governing law). They have 4,800 labeled clauses produced by three reviewers whose pairwise agreement is 81% overall but only 62% on two categories that are frequently confused.

780 words, about 4 min 4

Credit-memo summarization: high-risk approval in six weeks

A bank built a credit-memo summarization assistant on Granite-13b for its commercial lending analysts. The assistant condenses credit files of up to 200 pages into a two-page summary covering borrower background, facility structure, covenant package, and key risks.

869 words, about 4 min 5

Peak-hour timeouts on a two-node dedicated deployment

A customer-support assistant runs on a watsonx.ai dedicated deployment of Granite-13b with 2 GPU nodes. Traffic is 20 requests per minute for most of the day and spikes to 180 requests per minute every Monday at 09:00.

797 words, about 4 min