Image by Author | Canva
# Introduction
Personally, I find it amazing that computers can process language at all. It’s like watching a baby learn to talk, but in code and algorithms. It feels strange sometimes, but it’s exactly what makes natural language processing (NLP) so interesting. Can you actually make a computer understand your language? That’s the fun part. If this is your first time reading my fun project series, I just want to clarify that the goal here is to promote project-based learning by highlighting some of the best hands-on projects you can try, from simple ones to slightly advanced. In this article, I’ve picked five projects from major NLP areas so you can get a well-rounded sense of how things work, from the basics to more applied concepts. Some of these projects use specific architectures or models, and it helps if you understand their structure. So if you feel you need to brush up on certain concepts first, don’t worry, I’ve added some extra learning resources in the conclusion section 🙂
# 1. Building Tokenizers from Scratch
Project 1: How to Build a Bert WordPiece Tokenizer in Python and HuggingFace
Project 2: Let’s build the GPT Tokenizer
Text preprocessing is the first and most essential part of any NLP task. It’s what allows raw text to be converted into something a machine can actually process by breaking it down into smaller units like words, subwords, or even bytes. To get a good idea of how it works, I recommend checking out these two awesome projects. The first one walks you through building a BERT WordPiece tokenizer in Python using Hugging Face. It shows how words get split into smaller subword units, like adding “##” to mark parts of a word, which helps models like BERT handle rare or misspelled words by breaking them into familiar pieces. The second video, “Let’s Build the GPT Tokenizer” by Andrej Karpathy, is a bit long but such a GOLD resource. He goes through how GPT uses byte-level Byte Pair Encoding (BPE) to merge common byte sequences and handle text more flexibly, including spaces, punctuation, and even emojis. I really recommend watching that one if you want to see what’s actually happening when text gets turned into tokens. Once you get comfortable with tokenization, everything else in NLP becomes much clearer.
# 2. NER in Action: Recognizing Names, Dates, and Organizations
Project 1: Named Entity Recognition (NER) in Python: Pre-Trained & Custom Models
Project 2: Building an entity extraction model using BERT
Once you understand how text is represented, the next step is learning how to actually extract meaning from it. A great place to start is Named Entity Recognition (NER), which teaches a model to spot entities in a sentence. For example, in “Apple reached an all-time high stock price of 143 dollars this January,” a good NER system should pick out “Apple” as an organization, “143 dollars” as money, and “this January” as a date. The first video shows how to use pre-trained NER models with libraries like spaCy and Hugging Face Transformers. You’ll see how to input text, get predictions for entities, and even visualize them. The second video goes a step further, walking you through building an entity-extraction system by fine-tuning BERT yourself. Instead of relying on a ready-made library, you code the pipeline: tokenize text, align tokens with entity labels, fine-tune the model in PyTorch or TensorFlow, and then use it to tag new text. I’d recommend this as your second project because NER is one of those tasks that really makes NLP feel more practical. You start to see how machines can understand “who did what, when, and where.”
# 3. Text Classification: Predicting Sentiment with BERT
Project: Text Classification | Sentiment Analysis with BERT using huggingface, PyTorch and Python Tutorial
After learning how to represent text and extract entities, the next step is teaching models to assign labels to text, with sentiment analysis being a classic example. This is a pretty old project, and there’s one change you might need to make to get it running (check the comments on the video), but I still recommend it because it also explains how BERT works. If you’re not familiar with transformers yet, this is a good place to start. The project walks you through using a pretrained BERT model via Hugging Face to classify text like movie reviews, tweets, or product feedback. In the video, you see how to load a labeled dataset, preprocess the text, and fine-tune BERT to predict whether each example is positive, negative, or neutral. It’s a clear way to see how tokenization, model training, and evaluation all come together in one workflow.
# 4. Building Text Generation Models with RNNs & LSTMs
Project 1: Text Generation AI – Next Word Prediction in Python
Project 2: Text Generation with LSTM and Spell with Nabil Hassein
Sequence modeling is about tasks where the output is a sequence of text and it’s a big part of how modern language models work. These projects focus on text generation and predicting the next word, showing how a machine can learn to continue a sentence one word at a time. The first video walks you through building a simple recurrent neural network (RNN)-based language model that predicts the next word in a sequence. It’s a classic exercise that really shows how a model picks up patterns, grammar, and structure in text, which is what models like GPT do on a much larger scale. The second video uses a Long Short-Term Memory (LSTM) to generate coherent text from prose or code. You’ll see how the model feeds one word or character at a time, how to sample predictions, and even how tricks like temperature and beam search control the creativity of the generated text. These projects make it really clear that text generation isn’t magic, it’s all about chaining predictions in a smart way.
# 5. Building a Seq2Seq Machine Translation Model
Project: PyTorch Seq2Seq Tutorial for Machine Translation
The final project takes NLP beyond English and into real-world tasks focusing on machine translation. In this one you build an encoder-decoder network where one network reads and encodes the source sentence and another decodes it into the target language. This is basically what Google Translate and other translation services do. The tutorial also shows attention mechanisms so the decoder can focus on the right parts of the input and explains how to train on parallel text and evaluate translations with metrics like BLEU (Bilingual Evaluation Understudy) score. This project brings together everything you’ve learned so far in a practical NLP task. Even if you’ve used translation apps before, building a toy translator gives you a hands-on sense of how these systems actually work behind the scenes.
# Conclusion
That brings us to the end of the list. Each project covers one of the five major NLP areas: tokenization, information extraction, text classification, sequence modeling, and applied multilingual NLP. By trying them out, you’ll get a good sense of how NLP pipelines work from start to finish. If you found these projects helpful, give a thumbs-up to the tutorial creators and share what you made.
For learning more, the Stanford course CS224N: Natural Language Processing with Deep Learning is an excellent resource. And if you like learning through projects, you can also check out our other “5 Fun Projects” series:
Kanwal Mehreen is a machine learning engineer and a technical writer with a profound passion for data science and the intersection of AI with medicine. She co-authored the ebook “Maximizing Productivity with ChatGPT”. As a Google Generation Scholar 2022 for APAC, she champions diversity and academic excellence. She’s also recognized as a Teradata Diversity in Tech Scholar, Mitacs Globalink Research Scholar, and Harvard WeCode Scholar. Kanwal is an ardent advocate for change, having founded FEMCodes to empower women in STEM fields.