ML
ML Interview Roadmap
← Study plan·Deep Learning

Day 54 of 133

NLP practical tasks: NER, summarization, MT + DSA Graphs

BIO tagging vs span-based; ROUGE / BERTScore / LLM-as-judge; beam search.

DSA · NeetCode Graphs

  • Walls And GatesDSA · Graphs

    Interview questions to prep

    1. Is this BFS, DFS, or Union-Find? Defend the choice over the other two.
    2. Walk through complexity in terms of V and E. Where do those costs come from?
    3. How would you handle disconnected components, self-loops, or duplicate edges?
  • Course ScheduleDSA · Graphs

    Interview questions to prep

    1. Compare BFS-based topological sort (Kahn's) vs DFS with cycle detection.
    2. How does this extend to course-schedule-ii where you must output the order?

NLP · Practical tasks

  • Named-entity recognition: BIO tagging, span-basedDeep LearningHF

    Interview questions to prep

    1. Compare BIO tagging vs span-based NER models.
    2. How would you evaluate a NER model — token-level F1 vs entity-level F1, and which actually matters?
  • Extractive vs abstractive summarizationDeep LearningHF

    Interview questions to prep

    1. How would you evaluate a summarization model — ROUGE vs BERTScore vs LLM-as-judge?
    2. Compare extractive vs abstractive summarization — when is each the right tool?
  • Machine translation: BLEU, beam searchDeep LearningHF

    Interview questions to prep

    1. Why is beam search standard for translation but not for open-ended generation?
    2. What is BLEU actually measuring, and where does it mislead?
  • Sentiment analysis: baselines, transformers, and failure modesDeep LearningNeetCode ML

    Interview questions to prep

    1. Design a sentiment classifier starting from TF-IDF + logistic regression and then a transformer baseline.
    2. How would you handle negation, sarcasm, domain shift, and class imbalance in sentiment analysis?
    3. Which metric would you report for an imbalanced sentiment dataset and why?

References & further reading