AI/ML/NLP Research Track: Orientation

This orientation is your master guide for the full track from Python foundations to multilingual NLP research projects.

---

Track Vision

You are not only learning concepts. You are building a research-ready engineering profile with public artifacts.

Expected outputs by track completion:

• multiple notebooks and project repos
• three core ML projects
• one multilingual NLP capstone
• a documented research workflow and portfolio

---

Prerequisites

• basic Python scripting
• comfort with CLI and virtual environments
• willingness to ship weekly work publicly

You do not need:

• advanced math background
• prior transformer fine-tuning experience

---

Module Map

1. Kaggle + Python Workflow
2. NumPy, Pandas, Matplotlib, Jupyter
3. ML Fundamentals with scikit-learn
4. Three Core Projects (spam, sentiment, tabular)
5. NLP Foundations (tokenization, embeddings, transformers)
6. Hugging Face Workflow
7. Multilingual Assistant Capstone (Norwegian + Urdu)
8. Research Reading + Portfolio Publishing

---

Suggested Pace

• 6-8 hours/week minimum
• complete one meaningful code deliverable weekly
• publish one progress note weekly (GitHub/Kaggle/LinkedIn)

---

Project Sequence

• Project 1: spam detection
• Project 2: movie sentiment analysis
• Project 3: tabular prediction model
• Capstone: Norwegian + Urdu AI assistant

Each project should include:

• problem statement
• baseline model
• metric report
• error analysis
• next-step improvement plan

---

Research Skill Layer

Every week:

• read 1 paper/blog
• extract abstract/problem/method/results
• write one reproducibility note

This habit turns learners into research contributors over time.

---

Portfolio Requirements

Before considering the track complete:

• GitHub has organized repos with clean READMEs
• Kaggle has active notebooks and experiments
• LinkedIn has technical build updates

---

FAQ

Can I skip Python foundations and jump to transformers?
Not recommended. You will struggle with debugging and data workflows.

Do I need paid tools?
No. Colab, Kaggle, Hugging Face, and GitHub are enough to start strongly.