Hacker News AIT2·72 pts
- Coding agents let a developer implement LLM inference on an FPGA without deep hardware background
- Lower barriers to low-level hardware work may widen the space of explorable technical directions
- The workflow depends on a large token budget and well-targeted problems