Applied Natural Language Processing (NLP) From Text Data to Intelligent Applications

Course Structure

• Duration: 4 Weeks (Onsite | Online) 

• Onsite Classes: Saturdays 10am - 5pm  (GMT+1) + Weekday Follow-up Practice Sessions 

• Online: Tuesdays & Fridays 5.30pm - 8.00pm  (GMT+1) | Sundays 4.30pm - 8.00pm (GMT+1) 

• Access to the Catalyst AI Hub Africa Alumni Network

• 30 days of post-programme support

• Labs: 12 Hands-On Visualisation Labs

• Projects: 3+ — Including Executive Data Story Presentation

• Theory vs Practical: 20% / 80%

Ideal For

This course is designed for:

• Data analysts and BI professionals who want to elevate the quality and impact of their visual outputs

• Data scientists who can build models but want to communicate findings more effectively

• Product managers and business analysts who present data to executives and clients

• Financial analysts and economists wanting to move beyond Excel charts into Python visualisation

• Researchers and academics needing publication-quality charts and figures

• Marketing and growth professionals working with campaign data and customer analytics

Course Description

Participants will learn to preprocess and engineer features from raw text, build classical and deep learning NLP models, implement sentiment analysis, text classification, and named entity recognition systems, work with transformer models including BERT, and deploy production-ready NLP applications — all applied to Nigerian and African language datasets and business scenarios.

Recommended prior knowledge: Comfortable with Python — functions, loops, classes, and basic scripting. Familiarity with Pandas and NumPy is helpful. No prior NLP or deep learning experience required.

Summary of Learning  

By the end of this course, participants will be able to:

• Build complete text cleaning and preprocessing pipelines — handling Nigerian and African text including Pidgin English, code-switching, multilingual content, and non-standard spelling — transforming raw noisy text into structured machine-readable representations

• Apply classical and modern text representation techniques — including TF-IDF, Word2Vec, GloVe, FastText, and sentence embeddings — and select the right representation strategy for a given NLP task

• Build, train, and evaluate multi-class text classification systems using both classical machine learning models and deep learning approaches — including handling class imbalance and producing fully documented performance reports

• Implement lexicon-based, machine learning, and aspect-based sentiment analysis systems — applied to Nigerian social media, customer feedback, and news text with cultural and linguistic context built into the approach

• Design and train custom Named Entity Recognition (NER) models using spaCy — extracting Nigerian-specific entities including companies, executives, financial figures, and regulatory bodies from domain text

• Build deep learning NLP models using bidirectional LSTMs and sequence-to-sequence architectures — and explain the architectural trade-offs compared to transformer-based approaches

• Fine-tune pre-trained BERT models on custom Nigerian NLP tasks — including text classification, named entity recognition, and question answering — achieving high accuracy with limited labelled training data

• Apply multilingual transformer models including mBERT and XLM-RoBERTa to process Yoruba, Hausa, Igbo, and other African languages with limited training resources

• Design production NLP system architectures — including text search, semantic similarity, document summarisation, and chatbot pipelines — with appropriate latency, scalability, and reliability considerations

• Build and deploy a complete NLP application as a production REST API — wrapping a full NLP pipeline in FastAPI, containerising with Docker, and deploying to a live cloud endpoint with monitoring and maintenance guidance