Error Correction Code Engineer

Project

Work done for a product or program.

• Design LDPC code constructions, parity check matrices and code rates matched to the measured raw bit error rate of the target NAND generation.
• Model decoder performance including waterfall and error-floor behaviour using density evolution and Monte Carlo simulation over measured channel data.
• Architect hard-decision and soft-decision decoder hardware, including layered min-sum scheduling, quantisation and throughput/area trade-offs.
• Define the read-retry and soft-information escalation ladder jointly with firmware so decoding effort scales with block age.
• Design DRAM on-die ECC, link ECC and system-level ECC schemes such as SEC-DED, Chipkill-style symbol codes and bounded-fault correction.
• Extract channel models from NAND threshold-voltage distribution data and cycling/retention stress data to drive code design.
• Implement and verify RTL for encoder and decoder engines, closing throughput at the required gigabytes per second and latency budget.
• Quantify uncorrectable bit error rate and validate against product reliability targets such as 1e-16 UBER.

Function

Work that keeps the discipline running: methods, tools, suppliers and people.

• Design die-level and plane-level redundancy schemes (RAIN/RAID) and model their capacity and performance overhead.
• Collaborate with standards efforts and publish or patent coding innovations relevant to memory channels.

Project

1. Produce and hand over the role's key deliverables
   • LDPC code design with parity check matrix and rate plan
   • Decoder performance simulation reports including error-floor analysis
   • Encoder and decoder RTL with throughput and area closure
   • Channel model extracted from NAND characterisation data
   • Read-retry and soft-decode escalation specification
   • UBER validation report against product reliability targets
2. Design the blockGeneral
   • Write or capture the design to the specification
   • Run lint, simulation or circuit checks
   • Close design reviews and their actions
3. Support integration and sign-offGeneral
   • Fix bugs found by verification
   • Support timing and physical closure
   • Deliver handoff files and documentation

Function

1. Keep design methods currentGeneral
   • Maintain coding and design standards
   • Build and maintain reusable blocks
2. Develop peopleGeneral
   • Mentor newer engineers
   • Run design reviews and knowledge sharing sessions