Machine learning & data science
Solve NLP tasks (classification, NER, similarity)
Tokenization, transformers via Hugging Face, fine-tune small encoders for classification/NER.
~20 focused hours·intermediate
Tools: Hugging Face Transformers, spaCy, tokenizers, scikit-learn
Market relevance — share of job ads asking for this
Prerequisite capability — not asked for directly, but needed for others.
What employers mean
You should be able to…
- Tokenize and preprocess text correctly for a transformer model (truncation, padding, special tokens)
- Fine-tune a small pretrained encoder (BERT/DistilBERT) for text classification
- Build a Named Entity Recognition (NER) model for a domain-specific entity set
- Compute text similarity/embeddings for search or deduplication
- Handle multilingual or code-mixed text (e.g. Hindi, Hinglish) preprocessing challenges
- Evaluate classification/NER with the right metrics (F1, precision/recall per class, not just accuracy)
- Know when a classic ML baseline (TF-IDF + logistic regression) beats a transformer for a small dataset
Needs first: Train and evaluate classical ML models
Learn — free, link-checked
The few resources that matter
Course · beginner · 60 min · huggingface.co
Introduction
The standard free path into transformers -- pipelines, tokenizers and the Transformer architecture from the team that builds the library. — Hugging Face
Read · intermediate · 40 min · huggingface.co
Text classification
Copy-correct recipe for fine-tuning a small encoder (e.g. DistilBERT) for classification -- directly reusable for a support-ticket or sentiment model. — Hugging Face
Read · intermediate · 40 min · huggingface.co
Token classification
Official NER fine-tuning recipe with the BIO-tagging and evaluation details classical-NLP interview questions probe for. — Hugging Face
Practice
Hindi/Hinglish support-ticket classifier and entity extractor
Collect or synthesize a small labeled dataset of Hindi/Hinglish customer support tickets. Build a TF-IDF baseline classifier, then fine-tune a small multilingual encoder (e.g. mBERT or IndicBERT) for the same classification task and compare. Additionally extract key entities (product name, order ID) with a NER model or rule-based fallback.
Done when
- Labeled dataset (100+ examples) with a documented labeling scheme and train/test split
- Both TF-IDF baseline and fine-tuned transformer trained, with F1 per class compared in a table
- NER or entity extraction component demonstrated on at least 10 example tickets with correct/incorrect cases shown
- README honestly states which approach won and why, including any code-mixing failure cases found
Prove it
Evidence a recruiter can check
- Public GitHub repo with dataset (or synthetic generation script), training code and eval notebook
- Per-class F1 table comparing baseline vs transformer model
- Model card or short doc listing known failure modes on code-mixed input
- Can explain live the tokenization/preprocessing choices made for Hindi/Hinglish text
Interview
Questions you'll get asked
- How would you build a classifier to route support tickets into 5 categories, and what would you try first?
- Walk me through fine-tuning BERT for a classification task -- what changes vs training from scratch?
- How do you evaluate a NER model, and why isn't accuracy the right metric?
- What challenges come up with Hindi or Hinglish (code-mixed) text that don't come up in English?
- When would a TF-IDF + logistic regression baseline be a better choice than fine-tuning a transformer?
- How do you handle class imbalance in a multi-class text classification problem?
- What's the difference between a tokenizer's vocabulary and the model's embedding layer?