Manufacturing Test Data & Yield Pipelines
Wafer sort and final test on ATE, STDF files, bins, wafer maps and yield metrics, and the data pipeline that ingests, parses and warehouses test results at fab scale with PAT outliers and die-level traceability.
Overview
Why this topic
Every chip is tested at least twice before it ships: once while it is still part of the wafer, and again after it is packaged. Each test writes a record of every measurement for every die. A product line produces millions of those records a day, and they are how a company finds out that a process step drifted, that a test program has a bug, or that a customer return came from a known bad wafer.
The data engineer or infrastructure engineer on a test or product engineering team owns the path from the tester to the dashboard: collecting files from testers around the world, parsing STDF, storing it so a query over a year of lots returns in seconds, computing yield, flagging outliers and keeping every unit traceable to the wafer it came from.
Interview questions
- Design the data pipeline that takes STDF files from testers at several test houses into a warehouse that yield engineers can query. What are the hard parts?
- How do you compute yield correctly when parts are retested? Show the query.
- Explain PAT. How would you implement a dynamic PAT screen on a wafer, and what are the risks?
- A yield engineer says wafer maps for one product show a new failure pattern. How do you find the cause using the data you have?
- How do you make every shipped part traceable to its wafer, and how is that used when a customer returns a part?
- What would you put on a yield dashboard for a product line, and how do you alert on an excursion without crying wolf?
Sign up free to read the full topic
More AI Infrastructure topics
- GPU Fleet Health: XID Errors, ECC & DCGM
- GPU Node Bring-up & Burn-in
- GPU Quotas, Utilization & Chargeback On-prem
- GPU Virtualization: Passthrough, vGPU, SR-IOV & MIG in VMs
- GPUs on Kubernetes: GPU Operator, Device Plugin, MIG & Time-slicing
- LLM Inference: KV Cache, Batching & Scheduling
- Mixture of Experts (MoE) Inference
- Multi-tenant GPU Clusters & Confidential Computing
- Post-silicon Validation & Bring-up Labs
- Power & Cooling for GPU Racks
- PyTorch torch.compile, FSDP2 & Training Optimization
- Quantization & Model Compression
Start free on Cariara · Coding problems · System design studio