Suraj Singh

Distributed systems engineer, 13+ years. Currently building LLM inference infrastructure at Meta. Previously AWS OpenSearch.

Projects

• OpenSearch shard allocator: from weight factors to constraints — why I reversed my own design
• LLM inference at Meta — scheduling, KV-cache, latency at scale
• Killing the crash loop — 52 crashes to zero
• Cutting runtime by 80% — coming soon

Elsewhere

• LinkedIn
• Resume