$ whoami
Hi, I'm Advait.
I make complicated systems behave.
> Sometimes they even stay that way.
Senior SRE building resilient infrastructure, secure platforms, and distributed systems. From cloud architecture and PostgreSQL reliability to enterprise security appliances, I work where software meets production reality.
profile.card
- role
- Senior Site Reliability Engineer
- currently
- Defenxee · Pune, India
- also
- M.Tech in Cloud Computing, IIT Patna
- focus
- Kubernetes · PostgreSQL · AWS/Azure · Observability
kadvaitp@gmail.com
# anatomy of a request
How a Request Actually Travels.
An illustrative production path — network, application and data layers, with the observability plane watching all of them. Hover a component, trace a request, or break something on purpose.
drag the diagram sideways to see the full path →
// illustrative architecture and timings — a composite of patterns I've operated, not a live system or an employer deployment.
— · awaiting trace · simulated
# about
The Human Behind the Terminal.
I'm a Site Reliability Engineer who somehow ended up enjoying the things most developers try very hard to avoid: networking, databases, infrastructure failures, and the occasional 2 AM incident.
Over the last few years I've gone from owning backend APIs, to owning production infrastructure for ~140 enterprise customers, to owning the architecture and reliability of a 19-service security appliance. The common thread: being the person responsible when it breaks — and making sure it breaks less.
I work across the boundaries of DevOps, SRE, backend development, networking, security, and distributed systems. I care about understanding how systems actually operate, not just making them pass a deployment pipeline.
Currently also pursuing an M.Tech in Cloud Computing at IIT Patna, because apparently a full-time on-call brain needed homework.
"PostgreSQL babysitter. Kubernetes negotiator. Professional reader of logs that look personally offended."
things_i_believe.txt
- 01Uptime is temporary.
- 02Every system eventually becomes legacy.
- 03YAML errors are psychological warfare.
- 04Monitoring something doesn't automatically make it reliable.
- 05A backup that hasn't been restored is just optimism.
- 06The best incident response begins before the incident.
# experience
A Brief History of Keeping Things Running.
Production systems, complicated migrations, and lessons learned the hard way.
- redaction p95 latency (−82%)
- 705 → 129ms
- redaction p95 latency (−82%)
- Go tests maintained
- 1,307
- Go tests maintained
- services owned end to end
- 19
- services owned end to end
- MTTR
- −50%
- MTTR
- database incidents
- −40%
- database incidents
- infrastructure cost
- −20%
- infrastructure cost
- deployment failures
- −65%
- deployment failures
- REST APIs owned
- 250+
- REST APIs owned
- SQL queries optimized
- 400+
- SQL queries optimized
- customers served
- 100+
- customers served
May 2026 — Present
Senior Site Reliability Engineer
Defenxee Private Limited
Engineering ownership of a 19-service enterprise AI-security appliance — architecture, delivery and reliability — while managing a 7-person engineering team.
May 2023 — May 2026
DevOps Engineer
AssetCues Solutions Private Limited
Owned production availability, deployment and recovery for infrastructure supporting ~140 enterprise customers. Managed 6 DevOps engineers and worked alongside 20 developers.
Jan 2022 — May 2023
Backend Engineer — Go / PostgreSQL / AWS
Justlog Solutions
Owned 250+ REST APIs across 19 logistics domains for a platform serving 100+ customers, and migrated it from .NET to Go/PostgreSQL/AWS with API parity.
# projects
Things I've Built, Broken, and Fixed.
Selected systems, architectural decisions, and problems worth solving.
# stack
Tools of the Trade. And Occasionally Destruction.
Things I've used in production. Items marked with ✦ come with commentary.
Cloud & Infrastructure
- AWS
- Azure
- Kubernetes✦A distributed systems simulator with production consequences.
- Docker
- Terraform✦Because clicking through cloud consoles is not a deployment strategy.
- Helm
- Ansible
Programming
- Go
- Python
- Bash
- TypeScript
- SQL
- C#
- YAML✦Whitespace-based emotional damage.
Databases & Data
- PostgreSQL✦We've been through things together.
- MySQL
- Redis
Observability
- Prometheus✦Collecting evidence that something is about to go wrong.
- Grafana
- Loki
- OpenTelemetry
- Jaeger
- SigNoz
- Thanos
- Alertmanager
Networking & Security
- Nginx
- TLS
- ICAP
- HTTP/3
- QUIC
- TPROXY
- nftables
- LDAP/Kerberos
- RBAC
Delivery & Operations
- Linux✦Where the troubleshooting actually happens.
- GitHub Actions
- Jenkins
# philosophy
How I Think About Systems.
Reliability is designed, not wished into existence.
Recovery paths, failure modes and capacity limits are decided up front — or decided for you at 2 AM.
Observability should explain failures, not just announce them.
An alert that says 'something is wrong' is a smoke detector. Good telemetry tells you which room is on fire and why.
Automation should remove uncertainty, not create new mysteries.
If nobody understands what the pipeline does, it's not automation, it's folklore with a cron schedule.
Simple operational models beat clever complexity.
The system you can reason about during an incident is worth more than the one that wins architecture debates.
# beyond engineering
There's a Human Outside the Terminal. Allegedly.
01 /
Two wheels, fewer alerts.
I ride motorcycles. It's the one place where nobody can page me and the only latency that matters is reaction time.
02 /
Reading things that don't have stack traces.
I read widely and enjoy exploring ideas well outside day-to-day infrastructure work.
03 /
Leadership and learning.
I enjoy mentoring, teaching technical concepts, working closely with engineers, and understanding systems at a deeper level than the docs admit.
a detour, once
Recommended for NDA Course 144 (Air Force — Flying), All India Rank 411. A different path I didn't take — but the discipline it demanded stuck around.
# writing
From the Engineering Journal.
Notes from Production — things I've built, broken, debugged, and occasionally understood.
Personal SRE Dashboard
last 24h · auto-refresh: noCoffee dependency
Critical
Open incidents
0 (for now)
YAML tolerance
Degraded
Sleep schedule
Best effort
Production anxiety
Monitored
// simulated metrics. actual human reliability may vary.
# contact
Have an Interesting Problem?
Let's build something that doesn't wake anyone up at 2 AM.
I'm always interested in conversations about distributed systems, SRE, platform engineering, cloud infrastructure, security, and difficult engineering problems.
kadvaitp@gmail.com