Hardware-Adjacent Project Deep Dive

Project deep-dive rounds can focus heavily on prior HPC / GPU / hardware-adjacent projects. Candidates were asked to walk through CV projects, LSF CPU / GPU clusters, DCGM, workflow analysis for GPU utilization, OpenTelemetry, hardware-design compute needs, and Python / Hive / Iceberg libraries they had previously built or used. Requirements Prepare to walk through prior projects at implementation depth: What the project did and why it mattered. Which technologies, libraries, and infrastructure were used. How the system interacted with hardware, GPU clusters, or hardware-design workflows. How GPU utilization or compute-resource needs were measured. How OpenTelemetry-style instrumentation or monitoring fit into the system. How internally developed Python libraries were designed, deployed, and maintained. How data moved through Hive, Iceberg, or similar table formats. Notes This is not a generic resume skim. Interviewers can spend the full round on prior work and ask for exact technical details. For HPC / GPU work, be ready to explain cluster manager concepts, CPU vs GPU resource accounting, DCGM-style metrics, workflow-level utilization, and why those measurements matter for capacity planning. For data-platform work, be ready to describe schema evolution, table layout, partitioning, compaction, backfills, and deployment of Python modules used by other teams. Preparation Pick two projects and write a one-page technical brief for each: goal, architecture, hard bug, scale, metrics, and trade-offs. Prepare a diagram for any GPU / HPC cluster project showing scheduler, workers, metrics, storage, and users. Practice answering follow-ups at the level of API, data schema, deployment path, and operational failure modes.