InterviewsVector
Capability spine

Arc 10 · Staff AI Engineer

Inference, Training, and AI Infrastructure

Connect model execution to hardware reality: memory movement, precision, batching, parallelism, schedulers, storage, and failure recovery.

Exit capability: Capacity-plan and operate training and inference systems under real constraints.

Mapped lessons
7
Published now
0
Full-arc estimate
19 hours
Last edited
2026-08-11

Lesson sequence

Live units open into complete labs. Planned units stay visible to show the dependency path, but intentionally have no detail route.

  1. 01

    The Accelerator Memory Hierarchy

    Planned

    Reason from registers, SRAM, HBM, interconnect, and host memory to wall-clock performance.

    Concept labAdvanced110 min estimate

  2. 02

    Precision and Quantization as Error Budgets

    Planned

    Compare numeric formats and quantization methods through memory, throughput, calibration, and quality loss.

    Failure labAdvanced120 min estimate

  3. 03

    Continuous Batching and Queueing

    Planned

    Model prefill, decode, scheduling, tail latency, and throughput under mixed request shapes.

    Systems labAdvanced120 min estimate

  4. 04

    Parallelism for Inference

    Planned

    Choose tensor, pipeline, expert, and data parallelism from model shape and interconnect constraints.

    Design reviewAdvanced120 min estimate

  5. 05

    Distributed Training Without Topology Amnesia

    Planned

    Map optimizer state, gradients, activations, parameters, and collectives onto an actual cluster.

    Systems labAdvanced135 min estimate

  6. 06

    The Model-Serving Control Plane

    Planned

    Design placement, rollout, autoscaling, health, model loading, and rollback independent of request execution.

    Systems labAdvanced115 min estimate

  7. 07

    Load-Test an AI Service

    Planned

    Generate representative sequence shapes, concurrency, cancellations, and cache behavior while protecting quality signals.

    Build labAdvanced110 min estimateArtifact: AI load-test plan