ML
ML Interview Roadmap
← Study plan·MLOps

Day 58 of 133

Feature stores (Feast, Tecton) + point-in-time correctness + DSA Adv Graphs

Training/serving skew, offline/online stores, PIT joins.

DSA · NeetCode Advanced Graphs

  • Network Delay TimeDSA · Advanced Graphs

    Interview questions to prep

    1. Pick between Dijkstra, Bellman-Ford, Floyd-Warshall, MST (Prim/Kruskal), or topo sort — defend the choice.
    2. What does this problem assume about edge weights (non-negative? integer? bounded?) — and what breaks if those don't hold?
    3. Walk me through complexity in V and E, and the data-structure choice (heap vs Fibonacci heap vs array).
  • Swim IN Rising WaterDSA · Advanced Graphs

    Interview questions to prep

    1. Pick between Dijkstra, Bellman-Ford, Floyd-Warshall, MST (Prim/Kruskal), or topo sort — defend the choice.
    2. What does this problem assume about edge weights (non-negative? integer? bounded?) — and what breaks if those don't hold?
    3. Walk me through complexity in V and E, and the data-structure choice (heap vs Fibonacci heap vs array).

MLOps · Feature stores

  • Why feature stores: training/serving skew, reuseMLOpsTecton

    Interview questions to prep

    1. What is training/serving skew and how does a feature store eliminate it?
    2. When is a feature store overkill?
  • Feast, Tecton, Vertex Feature StoreMLOpsFeast

    Interview questions to prep

    1. Walk through how Feast separates the offline and online stores.
    2. Compare Feast vs Tecton vs Vertex Feature Store — when does each fit?
  • Point-in-time correctnessMLOpsTecton

    Interview questions to prep

    1. Why is point-in-time correctness critical for training data, and how does feature store handle it?
    2. Walk me through how a leak from forward-looking features actually breaks model rollout.

References & further reading