Ask HN: What things might help me to become inference engineer?
A former full-stack engineer, burned out on SaaS, asks the Hacker News community how to transition into inference engineering—a role in AI infrastructure—seeking advice from experienced AI Infra engineers on becoming a strong candidate in the field.
Background
- An **inference engineer** specializes in deploying and running trained AI models in production — making them fast, cost-effective, and reliable at scale. This is distinct from training models (which requires massive GPU clusters and data science skills); inference is about serving the model to users, optimizing latency, throughput, and hardware utilization.
- Hacker News (HN) is a tech community and link-aggregator run by Y Combinator. "Ask HN" posts are community questions. The audience is mostly software engineers, founders, and tech investors.
- The question reflects a growing career shift: as AI models become widely deployed, companies need engineers who understand low-level performance tuning, GPU programming (CUDA), memory management, batching strategies, and model compression (quantization, pruning), rather than just building web apps. This is a relatively new and specialized role that sits at the intersection of traditional backend/SRE work and ML ops.