Performance first.
I treat milliseconds as currency. At Dubicars, search dropped from 600ms to under 200ms; at SkillGigs, Mongo queries moved from minutes to milliseconds.
Read approachI treat milliseconds as currency. At Dubicars, search dropped from 600ms to under 200ms; at SkillGigs, Mongo queries moved from minutes to milliseconds.
Read approachKubernetes, serverless Lambdas, Redis, Elasticsearch — chosen on purpose. The Daraz price intel rebuild cut a 30-hour scrape down to eight.
Read approachTypo-tolerant search, multi-attribute filters, AI-generated listings. The backend disappears — the user just feels the speed.
Read approachWAFs, rate-limits, granular access control, OTP fail-over across providers. Hardening isn't a sprint — it's the floor I build on.
Read approachI migrate .NET to Node, integrate WhatsApp lead pipes, push iOS+Android notifications — and write the runbook the next on-call engineer reads.
Read approachThroughput, retries, idempotency, fail-over. Every system I ship assumes the network will lie, the provider will go down, and the data will arrive twice.
Indexes before rewrites. Cache before scale. I look at the query plan first, the architecture diagram second, and the team's calendar always.
Backend work earns trust by what it removes from the roadmap, not what it adds. I cut feature lists until they fit the runway and ship the spine first.
Show up. Do the rep. Read the logs again. Inspiration is what's left over when discipline finishes its sentence — most great systems are debugged on a Tuesday.
AI-generated listings (30%→70% trim accuracy), sub-200ms search, WhatsApp lead pipeline, multi-provider OTP fail-over.
View case study.NET→Node serverless migration, MongoDB query rewrites (mins→ms), WAF + rate-limit hardening, granular access control.
View case studyRe-architected Price Intelligence on Kubernetes — 30hr scrape → 8hr. Redis integrity, anti-ban throttling.
View case studyWordPress plugins + themes, Laravel + ReactJS builds, mockup-to-prod for multiple clients across markets.
View case studyMulti-attribute Elasticsearch with backward-compatible single search, typo-tolerant filters, Cloudflare edge caching.
View case studyCross-platform iOS + Android push at scale, structured logging, alerting and reporting baked into one observability layer.
View case studyTwo weeks of listening. Stakeholders, the codebase, the runbooks, the on-call. We surface what's actually slow before pretending we know.
One sentence the team can defend in a standup. The brief becomes a contract; the contract becomes a north star for every PR.
Schema, indexes, queues, cache layers — sketched against real traffic, not a whiteboard. We write the failing test first.
Node, AWS, Elastic, Mongo. Serverless where it pays, Kubernetes where it earns. Tested, observable, and small enough to read.
Launch is a checkpoint, not a finale. We measure, tune, document and hand over a system your team can actually keep on-call for.
“Talha rebuilt our scraping pipeline end-to-end — a four-times improvement, on a tighter footprint, without missing a release.”Engineering lead · Marketplace platform
“The kind of senior engineer who reads the query plan before opening Slack. We shipped faster, on a smaller team.”CTO · Remote SaaS
“No drama, no theatre. Just systems that scale, runbooks that work, and dashboards that finally tell the truth.”VP Engineering · E-commerce