Interview Stages

Round 1: Technical Discussion

• Process: Non-coding technical interview focused on ML/data topics.
• Topics:
  • How to determine if noisy data is usable for large models.
  • Methods for conducting ablation studies to evaluate data usefulness.
  • Follow-up on effective post-training techniques.
• Focus: Practical experience with data filtering and model training. Heavy emphasis on real-world knowledge.

Note: Felt like prior hands-on experience was critical—hard to improvise answers.