Skip to content

[email protected]:~$ watch career --source=resume

Aditya's résumé, as an observability platform

8 resolved incidents, 10 deploys, zero rollbacks — from the B.SQL turnaround after the ~70,000-order outage to the Datadog rollout and the $40M/yr FinOps cut. The postmortems are the career highlights; watch the golden signals; page the on-call if anything looks off.

regioneu-west-1 · Amsterdam on-call24/7 first shift2006-08-01 viewclassic →

Live console

guided terminal · type help
[email protected]:~$

how to read this dashboard

  • Incidents are the situations inherited at each job — the alert that was firing.
  • Postmortems record what was done and the business outcome.
  • SLOs measure the same claims as the classic résumé, in nines.
  • Deploys is the career timeline: every release shipped, none rolled back.

tip: ↑ / ↓ replays history

Incident feed — led, mitigated, resolved

all posts · 100% resolved
SEV1 INC-2022-042 resolved

B.SQL outage: ~70,000 orders lost, platform failing weekly

service booking.com/b.sql
opened 2022
ttr 6-week freeze → sustained turnaround
impact

A 2022 outage lost ~70,000 orders while the database platform failed several times a week — B.SQL was burning availability budget as a top offender.

detection

Availability-budget dashboards and the MySQL Availability Reporter: B.SQL sat at the top of the availability-budget-loss list per team.

resolution

Negotiated a 6-week feature freeze with senior management, then steadily improved the team’s processes. Introduced percentage-based canary deployments for database changes — a first for the team — plus risk-based test coverage, with a standing rule that every pull request must improve reliability or test coverage.

outcome

B.SQL taken off the top-offender list for availability budget loss (sustained since); on-call pages cut to roughly once a month.

Incident ManagementCanary DeploysAvailability Budget

SLOs & error budget

MySQL Availability Reporter · slo.yaml
— runbook coverage
reliabilityoff the top-offender list
sustainedsince the 2022 turnaround
on-call pages~1 / month
cost reduction65%+ · $40M+/yr
availability budget recovered 100%

B.SQL removed from the top offenders for availability budget loss per team — and kept off.

slotargetachievedbudget
availability leave the top-offender list off it — sustained since availability budget recovered
on-call health healthy rotation ~1 page / month Pythian handles L1
runbooks every failure mode ~95% coverage updated after every big failure
cost cut MySQL infra spend −$40M+/yr (65%+) Datadog licence ~50% of quote
compliance SOX · PCI · PII 100% no exceptions

Golden signals

metrics · 15s scrape
latency p99 ~15 ms
traffic 1M+
error rate (sim) ~0 · pages 1/mo
fleet saturation 20k+ servers

Latency, traffic, errors, saturation — the four golden signals, running on career-scale numbers: ~15ms p99 at 1M+ req/min across a 20,000+ server fleet, with runbooks covering ~95% of known failure modes and a pager that now fires about once a month.

Deploy timeline

release history · 0 rollbacks
  1. v2022.02 booking.com Feb 2022 → Present SRE Engineering Manager · Amsterdam B.SQL turnaround · Datadog rollout · −$40M/yr (65%+)
  2. v2019.08 deliveryhero/pricing Aug 2019 → Dec 2021 Engineering Manager · Berlin canary promoted: 1M+ req/min at ~15ms
  3. v2015.05 deliveryhero/orders May 2015 → Aug 2019 Senior Software Engineer · Berlin McDonald’s & Domino’s integrations · geolocation
  4. v2014.12 spaceways/backend Dec 2014 → Apr 2015 Senior Backend Developer · Berlin PHP (Phalcon) · Docker · Vagrant
  5. v2014.04 makemytrip/platform Apr 2014 → Nov 2014 Manager, Technology · Gurgaon CDN, deployment pipelines, live monitoring · +23% engagement
  6. v2012.03 fabfurnish/platform Mar 2012 → Apr 2014 Technical Architect / EM · Gurgaon MySQL + Solr: +30% efficiency · mobile traffic +40%
  7. v2010.10 belzubar/apps 2010 → 2012 Sr. Computer Scientist · New Delhi AWS management, architecture, database optimisation
  8. v2010.02 kaango/platform 2010 → 2010 Senior Software Developer · US (remote) two major platform features with the US team
  9. v2008.02 infoedge/naukri 2008 → 2010 Senior Software Engineer · Noida LAMP at scale · site speed +35%
  10. v2006.08 oracle/flexcube 2006 → 2008 Associate Consultant · Bengaluru initial commit: Flexcube core banking · PL/SQL

On-call

pager · pages ~1/month
on-call primaryAditya · ADS escalation rotation
escalation rotationADS-level · Manchester DB shifts · DML on-call (setting up)
runbooks~95% of failure modes
page frequency~1 / month
100% SOX100% PCI100% PII

Designed a healthy on-call model with an outsourced ops partner: Pythian handles routine L1, runbooks are updated after every new repeating or big failure, and onboarding comes with recorded training and shadowing. Leads response for large B.SQL incidents and takes regular on-call shifts.

Page the on-call (LinkedIn)