argakiig
distributed systems · production engineering · infrastructure & ai systems
i design and operate distributed systems where correctness, throughput, and cost all matter — schedulers, data pipelines, and the infrastructure underneath them.
lately i've been working at the intersection of high-performance compute and agent-assisted engineering — using ai tooling as a force multiplier on real production systems, not as a demo.
the thread through all of it: engineering value is moving away from implementation and toward judgment — what to build, where the system's boundaries are, and how it behaves when it breaks.
selected work
- 200+ TB of market data ingested and processed down to hundreds of GB, with strict chronological guarantees
- historical and streaming aggregation pipeline tested at thousands of transactions per second
- operated public rpc infrastructure and crypto node infrastructure across layer 1 and layer 2 — node operation, transaction pool tuning
- distributed schedulers and workers handling complex dependency graphs at scale
- observability and reliability engineering: backpressure, deduplication, recovery
- database-level deduplication and upsert patterns for high-volume ingest
- orchestration migrations from fire-and-forget to long-lived batched workers
- performance work across cpu and gpu — sieves, parallel pipelines, throughput tuning
- gpu rental fleet orchestration and cost-optimized batch compute
- local-first ai tooling, coding agents, and headless sdk integrations
- cloud infrastructure, kubernetes, and infrastructure as code for small teams
- technical documentation, style guides, and reusable agent skills