I'm Piyush β a backend infrastructure engineer based in Santa Clara, CA, currently building scalable, cloud-native platforms at Arista Networks. I spend my days deep in the guts of distributed systems β service meshes, container orchestration, network data planes, and the long tail of failure modes that come with them.
- π Working on β production-grade k8s + Istio platforms, and AI agents that triage incidents the way a senior on-call would
- π± Learning β eBPF dataplane acceleration, agentic workflows, and the open-source EDA toolchain
- βοΈ Writing β latest post: Why Traditional Alerting Is Broken β And How AI Agents Finally Fix It
- π Off-shift β poker (cash & MTT), long walks around the Bay, the occasional algo-trading rabbit hole
- π« Reach me β piyushbag.com Β· pbag@alumni.scu.edu
name: Piyush Bag
role: Software Engineer @ Arista Networks
location: Santa Clara, CA πΊπΈ
education: M.S. Computer Science, Santa Clara University
focus: backend Β· cloud-native Β· distributed systems Β· AIOps
interests: [k8s, service-mesh, observability, ai-agents, open-hardware]
side-quest: building tools for the open PCB / EDA ecosystem
motto: "debug the system, not the symptom"π
awesome-pcb-workflowis my current favorite hobby project β curating the full open-source PCB pipeline (schematic β SPICE β SI β layout β DFM β fab CI/CD β hardware test), OrCAD- and KiCad-friendly.
A few projects I've been pushing PRs and issues into lately:
Watch a snake nom my green squares. Powered by Platane/snk, regenerated daily by GitHub Actions.
| Theme | What & Why |
|---|---|
| π€ AI agents for infra | Replacing brittle threshold alerts with LLM-driven RCA loops β see the blog post |
| πΈοΈ Service mesh internals | Envoy filters, xDS, eBPF-accelerated paths, and where mesh actually earns its overhead |
| π Open hardware tooling | Code-first EDA (atopile, SKiDL), CI/CD for PCBs (KiBot), open SI/EMC simulation |
| π Poker math | Solver workflows, GTO ranges, and the eternal hunt for a +EV edge |
I'm always up for a chat about distributed systems, cloud-native platforms, AI-augmented ops, or open hardware β or a friendly cash game if you're in the Bay Area.


