Pfizer Externship via Extern·🟢 Open to workI train in practical IoT attack and defense techniques through a Hydroficient externship to build real skills and improve my resume.
Projects include the Hydroficient IoT Cyber Defense externship: launching realistic attacks on IoT infrastructure, implementing network defenses, and documenting a security playbook.
.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.
Developed and evaluated MLP neural networks to predict green vegetation, non-photosynthetic vegetation, and soil abundance fractions from hyperspectral reflectance data; processed a dataset of 1,723 samples with 210 spectral features, implementing full train/validation/test workflows in PyTorch; improved model Test R² from 0.871 to 0.896 through systematic architecture optimization and hyperparameter experimentation; visualized model performance using training/validation loss curves and true-vs-predicted analysis to evaluate generalization
I train in practical IoT attack and defense techniques through a Hydroficient externship to build real skills and improve my resume.
I am Roqaya El-rahwan, a first-year Computer Engineering student working to build hands-on experience in IoT security. I am completing the Hydroficient IoT Cyber Defense externship, where I practiced attacking and defending real IoT systems and documenting security playbooks to strengthen my resume.
Externships
Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship
In progress
Experience
Pfizer Extern
Pfizer · 2026 – Present
Document Intelligence Externship · 2026
Education
University of Toronto
Bachelor of Applied Science, Computer Engineering + PEY Co-op
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.
A prototype project within the Pfizer externship that built AI-powered document intelligence pipelines combining OCR, large language models, and retrieval-augmented generation. The work surveyed LLM fundamentals, designed end-to-end data extraction flows for PDFs, and produced prototype artifacts demonstrating automated document query and extraction.
I produced a technical write-up explaining how an LLM was trained, how tokenization and Transformer attention shape context weighting, and how iterative parameter updates improve next-token prediction when applied to pharmaceutical documents.
The submission explained how an LLM was trained on large text corpora, tokenised prompts, used Transformer attention to weight relevant context, and iteratively adjusted parameters during training to improve next-token predictions.
Developed and evaluated MLP neural networks to predict green vegetation, non-photosynthetic vegetation, and soil abundance fractions from hyperspectral reflectance data; processed a dataset of 1,723 samples with 210 spectral features, implementing full train/validation/test workflows in PyTorch; improved model Test R² from 0.871 to 0.896 through systematic architecture optimization and hyperparameter experimentation; visualized model performance using training/validation loss curves and true-vs-predicted analysis to evaluate generalization
Developed and evaluated MLP neural networks to predict green vegetation, non-photosynthetic vegetation, and soil abundance fractions from hyperspectral reflectance data
Engineered an end-to-end machine learning pipeline in Python to analyze structured historical Formula 1 race data; performed data cleaning, feature engineering, exploratory analysis, and model evaluation; implemented predictive algorithms to forecast race outcomes and visualized performance metrics using Matplotlib; structured codebase for modular experimentation and model comparison
Collaborated in an engineering team to design accessibility-focused trail infrastructure for a real client; translated stakeholder requirements into measurable technical constraints and evaluated multiple design alternatives; conducted feasibility analysis balancing environmental impact, cost, and usability considerations; produced structured design documentation and formal client presentations
Developed a React-based student support platform that centralizes U of T academic, career, financial, accessibility, and wellness resources into a single dashboard; built a multi-step onboarding system that collects students’ faculty, academic needs, challenges, and goals to generate personalized resource recommendations; implemented dynamic faculty-specific recommendations and direct routing to relevant university support services; designed an interactive end-to-end user experience including authentication, personalized onboarding, resource discovery, and dashboard interfaces
Developed a signal-processing workflow to clean and analyze high-dimensional brain imaging datasets; applied statistical analysis and dimensionality-reduction techniques (PCA) to extract meaningful patterns; generated data visualizations to support interpretation of neurological trends
Designed a custom 2-layer printed circuit board (PCB) in Altium Designer following engineering design requirements; created the schematic, selected and placed electronic components, defined the board outline, and routed PCB traces while meeting design constraints; applied PCB design principles including component placement, mounting holes, design rule checks (DRC), and manufacturable layout practices
