About

Harsh Bhuwania

Engineer who pulls the thread until it's deployable.

Harsh Bhuwania

Senior software engineer with about a decade of experience building large-scale infrastructure — first on networking platforms at Cisco, now on cloud infrastructure at Microsoft. I work where the cloud meets the metal: firmware, BMCs, telemetry, and the control planes that keep a global fleet healthy.

Started in platform software for IOS-XR at Cisco. Now at Microsoft on Azure Infrastructure, working on the software and firmware that keeps a global fleet of cloud hardware healthy. M.S. in Electrical Engineering from USC.

Based in Mountain View, CA. Reach me at hbhuwania@microsoft.com.

What I care about

Operational reality over slideware

I don't ship a design until it's deployable, measurable, and defendable. The weekly newsletters and one-pagers exist because the rest of the org needs to act on the same data I do.

Fix at source, not band-aids

I'd rather take an extra sprint to remove a class of bugs than ship the same patch a third time. I'll push back on whitelisting CVEs in favor of upstream patches with an actual closure plan.

Writer-first, sync when it matters

Long-lived artifacts — specs, wikis, runbooks — are my default; they outlast any meeting. But when the constraint space is tangled, I'd rather spend five minutes on a call than thirty in chat.

AI for the boring half of the job

A lot of my recent work has been building Copilots and bots that compress hours of log triage and CLI introspection into minutes, so the team can spend its energy on the interesting half.

What I'm good at

Firmware & embedded

  • Board bring-up
  • BMC software
  • OpenBMC / Yocto
  • Linux kernel-adjacent debugging
  • C / C++ → Rust migration

Reliability at fleet scale

  • Telemetry pipelines
  • KQL dashboards
  • Fault-code taxonomies
  • MTTR reduction
  • On-call DRI

Security as a program

  • Threat modeling
  • SBOM
  • FIPS / NIST alignment
  • Privilege separation
  • Platform Root of Trust

Performance & benchmarking

  • Empirical control-plane benchmarks
  • Memory & thread-pool root-causing
  • Benchmarks-in-CI

AI-assisted developer productivity

  • Log-analyzer Copilots
  • Introspection bots
  • AI-generated test cases

Architecture & leadership

  • Multi-semester roadmaps
  • Hardware program specs
  • Cross-org v-team coordination
  • External vendor integration

Languages & tools

  • C
  • C++
  • Rust
  • Python
  • Bash
  • KQL
  • Linux
  • Distributed systems

How I work

  • Convert ambiguity into bounded artifacts so others can execute without me.
  • Lean technical, operational, and strategic — often all three at once.
  • Comfortable in off-hours collaboration across time zones.
  • Equally happy as a deep diver or a cross-cutter, depending on what the problem needs.
  • Currently growing: delegating review load and institutionalizing dependency closure across teams.

Fun facts

  • Read 50 library books in the last 12 months — from Shōgun to Why We Sleep.
  • Take long walks with a podcast queue spanning sports, economics, psychology, and AI.
  • Believe the best debugging tool is a fresh pair of eyes — usually mine, after a walk.
  • Equally happy reading a Morgan Housel essay or a Keigo Higashino mystery.
  • Started in firmware at Cisco, now wrangling cloud infrastructure at Microsoft — still chasing the same kind of low-level puzzle.