Ali Abyer Nasir, engineer and tech lead at Wanile Technologies, Lahore

The road ahead is fog.I keep walking.

  • Engineerانجینئر
  • Architectمعمار
  • Operatorآپریٹر

IApproach

I ship production systems, not prototypes.

Four years of shipping things that have to hold.

I’m a full-stack engineer, AI engineer, and cloud architect, and the tech lead at Wanile Technologies. I’ve shipped production SaaS for fifty thousand users: the screen you touch, the APIs behind it, and the models doing the work.

I care about the unglamorous parts. Voice agents that answer in under 300ms. Retrieval that cites its sources. Cloud that stays up at 3am. Systems that are resilient, observable, and maintainable, then get out of the way.

  • Voice AI & conversational agents
  • RAG & LLM orchestration
  • Cloud infrastructure at scale
  • Real-time & 3D on the web
  • Production SaaS architecture
0+Users servedacross production SaaS
<0msVoice round-tripPanda Voice, in production
0+Concurrent SaaSshipped at Wanile
0Daily active visitsKitabh platform scale

IIThe work

Three I built to last.

Each one is still in production. That was the hard part.

01Conversational voice AI · SaaS

Panda Voice

A voice survey platform on WebRTC. Agents run structured Q&A or open conversation, backed by a RAG pipeline. Every call gets mined for sentiment, complaints, and insight.

Keeping voice round-trip under 300ms while injecting RAG context across multi-mode agent state. Edge-optimized WebRTC signalling with pre-warmed model connections.

  • Next.js 15
  • NestJS 11
  • ElevenLabs
  • OpenAI
  • Pinecone
  • WebRTC
  • Redis
Production · B2B
02Cloud infrastructure · 50K users

Kitabh

An Arabic-first AI publishing platform with 50,000+ readers. I owned the cloud: dual ASGs behind Cloudflare and an ALB, zero-downtime CI/CD, and a Bahrain to Frankfurt region migration with rollback documented at every step.

Zero-downtime deploys via CodePipeline → SSM → versioned S3 artifacts → ASG Instance Refresh at 100% min-healthy. Rollback time cut from ~15 minutes to under two.

  • AWS EC2 / ASG / ALB
  • CodePipeline
  • CloudWatch
  • Cloudflare
  • MongoDB Atlas
  • PM2
03Serverless HRMS · event-driven AWS

SmartHR

A fully serverless HRMS on AWS. Payroll, attendance, leave, and role-based access run as event-driven Lambdas — API Gateway into EventBridge, fanned out across functions in parallel, writing to DynamoDB and buffered through SQS. Infrastructure defined and shipped as code with SST.

Correctness under at-least-once delivery: making every handler idempotent so retries and parallel invocations never double-pay or double-count. Dedupe keys, conditional writes, and careful event ordering keep the ledger exact under concurrency.

  • AWS Lambda
  • EventBridge
  • DynamoDB
  • SQS
  • API Gateway
  • SST
Production · Serverless

Also on the bench

IIIInstruments

What I reach for.

Nothing listed here that hasn’t carried production weight.

AI & ML

  • Voice AI (ElevenLabs)expertcarried weight in — panda-voice · dual-mode agents
  • RAG / Pineconeexpertcarried weight in — panda-voice · pdf → embed → retrieve
  • LLM orchestrationexpertcarried weight in — panda-voice · model-per-task
  • WebRTC voiceexpertcarried weight in — panda-voice · voice-agent-demo
  • Whisper / TTSadvancedcarried weight in — kitabh narration · voice clone

Frontend

  • React 19expertcarried weight in — every major frontend
  • Next.js 15 / 16expertcarried weight in — kitabh · panda-voice · benchly
  • TypeScriptexpertcarried weight in — every production repo, no exceptions
  • Three.js / R3Fadvancedcarried weight in — panda-voice · interactive scenes
  • Tailwind v4expertcarried weight in — every interface since v4

Backend

  • NestJS 11expertcarried weight in — panda-voice · sengira
  • Express.jsexpertcarried weight in — kitabh api · benchly
  • MongoDBexpertcarried weight in — kitabh · panda-voice
  • PostgreSQLadvancedcarried weight in — sengira · benchly
  • Redis / BullMQadvancedcarried weight in — panda-voice cache · sengira queues

Cloud & DevOps

  • AWS EC2 / ASG / ALBexpertcarried weight in — kitabh v19 · live
  • AWS Lambda / Serverlessexpertcarried weight in — smarthr · event-driven · idempotent
  • CodePipeline CI/CDexpertcarried weight in — kitabh zero-downtime deploys
  • CloudWatchexpertcarried weight in — kitabh observability
  • Dockeradvancedcarried weight in — sengira local stack
  • k6 load testingadvancedcarried weight in — kitabh · 1500 vus through cloudflare

expert advanced

IVRecord

Where the time went.

Four years, three rooms, each one deeper than the last.

  • Mar 2021 – Feb 2024

    Zeen Technologies

    Full-Stack Engineer

    My first role shipping to real users. I built end-to-end systems, ran them on cloud under live traffic, and learned what production responsibility actually costs.

    • 2 production apps
    • 10K+ users
    • First cloud (GCP)
  • Feb 2024 – 2025

    SpyreSync

    Senior Engineer · Tech Lead

    I stopped writing code alone and started leading it. Broke legacy monoliths into event-driven serverless microservices on AWS. Owned architecture decisions, reviews, and planning.

    • 100% serverless
    • 5 microservices
    • SST IaC
  • 2025 – now

    Wanile Technologies

    Tech Lead · AI & Cloud Engineer

    Leading engineering across five SaaS products at once. Building sub-300ms voice agents, RAG pipelines that cite their sources, and cloud infra that deploys without downtime.

    • 50K+ users
    • <300ms voice
    • 5 SaaS products

VSignal

If this resonated,
get in touch.

Somewhere in Lahore, a terminal is listening.

Open to senior engineering roles & technical co-founder conversations.

31.5204° N  74.3587° ELahore  PKTBuilt with Next.js, Three.js, and a lot of fog