🟢 Open to work
I am currently testing different roles and career paths while studying Industrial Engineering and learning applied AI through a Pfizer externship.
Work shown includes learning-focused projects and the Pfizer Advanced: AI-Powered Document Insights & Data Extraction externship, highlighting practical skills I built while exploring roles.
.jpg)
Pfizer · ✅ Verified by Extern · ⏱️ In progress
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
I am currently testing different roles and career paths while studying Industrial Engineering and learning applied AI through a Pfizer externship.
I'm Elizabeth Saku, currently studying Engineering and Management for my bachelor's degree. I recently completed the Pfizer Advanced externship which focused on AI-Powered Document Insights and Data Extraction, where I gained hands-on experience in applying AI to solve real-world challenges.
Externships
Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship
Pfizer
Education
Technische Hochschule Ingolstadt
Engineering and Management · Class of 2028
Skills
✅ Verified by Extern · ⏱️ In progress
Prototype AI-powered document intelligence with Pfizer—using OCR, LLMs, and RAG to automate real enterprise PDF workflows and build a standout portfolio project.
This externship prototyped AI-powered document intelligence for enterprise PDF workflows, combining OCR, large language models, and retrieval-augmented generation. The work surveyed LLM fundamentals, mapping model components and inference patterns, and documented integration points with OCR and RAG approaches.
I produced a written submission that explained LLM mechanics, showing how tokenization, Transformer context modeling, and iterative next-token prediction enable models to process pharmaceutical documents.
The submission explained LLMs as models trained on large text corpora that tokenize input, use Transformer attention to model context, and generate answers by predicting next tokens iteratively, framing them as advanced autocomplete rather than retrieval systems.