I'm a Digital Technologies (AI) student eager to learn, grow, and build practical AI skills. Exploring the field while preparing for a career in artificial intelligence.
Digital Technologies student building academic and externship projects in Python, machine learning, data analysis, and AI, with a focus on practical problem-solving.
University Centre David Game London — Academic Project · Dec 2025
An academic machine-learning project investigating whether mushroom characteristics could be used to classify specimens as edible or poisonous. The project combined data preparation, exploratory analysis, model development, and evaluation using Excel, Python, and Google Colab.
University Centre David Game London — Academic Project · Jul 2026
An academic business-intelligence project examining public footfall data from Leeds city centre to understand movement patterns and assess potential retail locations. The project focused on turning raw location data into clear, decision-relevant insights.
University Centre David Game London — Academic Project · Dec 2025
An academic systems-analysis project based on an interview with a real business, an independent gluten-free bakery in London. I investigated its manual ordering, stock-control, and customer-management processes and proposed two connected digital systems.
.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'm a Digital Technologies (AI) student eager to learn, grow, and build practical AI skills. Exploring the field while preparing for a career in artificial intelligence.
Externships
Pfizer Advanced: AI-Powered Document Insights & Data Extraction Externship
Pfizer
Education
University Centre David Game London
Digital Technologies (AI) · Class of 2026
University Centre David Game London — Academic Project · Dec 2025
An academic machine-learning project investigating whether mushroom characteristics could be used to classify specimens as edible or poisonous. The project combined data preparation, exploratory analysis, model development, and evaluation using Excel, Python, and Google Colab.
An academic machine-learning project investigating whether mushroom characteristics could be used to classify specimens as edible or poisonous. The project combined data preparation, exploratory analysis, model development, and evaluation using Excel, Python, and Google Colab.
I cleaned and prepared more than 61,000 records, encoded categorical data, examined feature relationships, and trained and compared multiple classification models, including Random Forest, SVM, KNN, logistic regression, decision trees, and ensemble methods.
The strongest models achieved near-perfect classification performance, with Random Forest reaching 100% test accuracy and a stacking ensemble reaching 99.99%. The analysis identified odour, gill colour, and spore-print colour as the most influential features.
University Centre David Game London — Academic Project · Jul 2026
An academic business-intelligence project examining public footfall data from Leeds city centre to understand movement patterns and assess potential retail locations. The project focused on turning raw location data into clear, decision-relevant insights.
An academic business-intelligence project examining public footfall data from Leeds city centre to understand movement patterns and assess potential retail locations. The project focused on turning raw location data into clear, decision-relevant insights.
I sourced and prepared public data, compared footfall across locations and time periods, identified patterns and inconsistencies, and designed a dashboard mock-up with relevant KPIs and visualisations. I also evaluated data quality, limitations, and business requirements.
The analysis highlighted differences in activity levels and consistency between city-centre locations, demonstrating how footfall data could support retail site selection. The final report presented practical recommendations and a proposed dashboard design for faster interpretation.
University Centre David Game London — Academic Project · Dec 2025
An academic systems-analysis project based on an interview with a real business, an independent gluten-free bakery in London. I investigated its manual ordering, stock-control, and customer-management processes and proposed two connected digital systems.
An academic systems-analysis project based on an interview with a real business, an independent gluten-free bakery in London. I investigated its manual ordering, stock-control, and customer-management processes and proposed two connected digital systems.
I gathered requirements through an interview and workflow observation, mapped stakeholders and processes, and designed a proposed POS and inventory system alongside a customer-ordering and loyalty web application. I also produced a traceability matrix, process flows, and state diagrams.
The project produced a detailed, paper-based software proposal using a hybrid Waterfall and Agile lifecycle. It covered feasibility, costs, scheduling, security, testing, and implementation, showing how the bakery could reduce manual errors, improve stock visibility, and strengthen customer engagement.
✅ 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.
The project prototyped AI-powered document intelligence for enterprise PDF workflows using OCR, large language models, and retrieval-augmented generation. Work included researching model architectures and data flows, designing end-to-end processing steps, and producing explanatory deliverables about LLM operation.
I produced a detailed explanation of LLMs that covered training on large text corpora, the Transformer architecture for contextual prediction, and post-training fine-tuning with human feedback to improve accuracy and safety.
The submission explained LLMs by describing their training on large text corpora, the Transformer architecture for contextual word prediction, and post‑training fine-tuning with human feedback to improve accuracy and safety.
The module work summed multiple Colab notebooks where I loaded datasets, cleaned CSV/JSON/text, wrote Python functions and control flow, and applied image preprocessing for OCR. Deliverables included runnable Colab notebooks and a cleaned document image.
In the Extract & Structure Data from SDFs task I ran PyMuPDF on a three-page SDF, inspected extracted text and bounding boxes, and documented successes and issues. I reported that most text (product info, dates, lot numbers) was captured, noted table and multi-word splitting problems, and…
The project processed a scanned Certificate of Quality, ran three OCR engines, cleaned outputs with regex, and compared text accuracy and layout preservation. The work found PaddleOCR offered the best balance of layout-aware structure, with Tesseract as a strong text fallback.