Fixed price against a written scope, in two or three milestones. You know the total before I start, and it does not move unless you change the scope in writing. Retainers are billed monthly in advance.

Is the AI work real, or a wrapper around an API?

Calling a model is the easy part. The work is everything around it: latency budgets, streaming, retries when a provider degrades, guardrails on what the agent can act on, transcript storage, and the integration write backs that make the output useful. I have shipped a voice agent into live outbound calling, which is an unforgiving place to learn that difference.

What if it breaks after handover?

Thirty days of support is included in every build. If something I wrote misbehaves, I fix it at no cost. Beyond that most clients move onto a small retainer, but there is no obligation and you keep all the code either way.

Entirely. It ships to your repository, on your infrastructure, with documentation written for whoever comes next. No proprietary wrapper, no hosting lock in, nothing that requires me to stay involved.

Timezone and availability?

I work from Lahore at UTC+5, which overlaps the UK and EU working day comfortably and catches the US East Coast morning. Most of my client work has been for UK businesses. I reply to email within 24 hours, usually sooner.

Are you available for full time roles?

Yes. My last engagement ended in August 2026 and I am open to full stack, backend or AI engineering roles, remote or onsite in Lahore. My résumé has the long version, and the about page covers what I am looking for.

Full Stack AI Engineer · Lahore, Pakistan

SYSTEMSTHAT RUNTHEMSELVES.

Five years building full stack products and production AI systems. I ship voice and LLM agents, MCP servers, and the integration plumbing underneath them, then deploy the whole thing on AWS and keep it standing.

Start a project →See the work →

Available now for freelance builds and full time roles. Last engagement ended August 2026.

Specification

Discipline: Full Stack AI Engineer
Experience: 5+ years in production
Core stack: Python, Next.js, PostgreSQL, AWS
AI: OpenAI, Retell AI, MCP servers
Based: Lahore, PK · UTC+5
Replies in: Under 24 hours
Status: Open for work

AI Agents.

Voice, chat and MCP servers

Integrations.

Systems that stop disagreeing

Automation.

The work nobody should be doing

Cloud.

AWS, deployed and kept up

The toolkit

Things I use in production, not a list of everything I have read about

AI engineering

OpenAI, Retell AI, LLM integration, conversational voice agents, prompt engineering, and MCP servers built and deployed for agent tooling.

AI assisted development

Claude Code, Codex and Cursor are in my daily setup. They make the mechanical work faster. Judgement, architecture and review stay mine.

Languages & frameworks

Python with FastAPI, Django and Flask. PHP with Laravel. TypeScript, React, Next.js and Tailwind on the front.

Backend & APIs

REST API design, webhooks, microservices, queue workers. Twilio, Shopify, eBay and Royal Mail integrations in production.

Databases

PostgreSQL, MySQL, MongoDB, Redis and Supabase, including the query tuning that keeps them fast under load.

Cloud & DevOps

AWS across EC2, ECS, RDS, ElastiCache, VPC and Secrets Manager. Docker, Nginx, Linux VPS and Git.

What I actually do

Six things done properly, rather than twenty listed

AI agents in production

Voice and chat agents that do real work: conversation handling, intent routing, transcript capture, CRM write back. Built to survive live traffic, not a demo.

MCP servers & agent tooling

Exposing your internal systems to LLM agents safely through Model Context Protocol servers. Built, integrated and deployed, with the permission boundaries thought through first.

API design & integration

Two systems that were never meant to speak: storefront, ERP, courier, payments, CRM. Versioned, documented, idempotent.

Automation workflows

The task someone does every morning at nine, replaced by a schedule, a queue and an alert for when it genuinely needs a human.

Full stack product builds

Architecture through to interface: Python services, Next.js front ends, Postgres underneath, shipped as one coherent product rather than handed between people.

Cloud deployment & hardening

AWS deployment done properly: ECS, RDS, ElastiCache, private networking, secrets kept out of the repository. Then made boring enough to forget about.

Selected work

Tap to expand

Problem

Reps spent the first half of every day on outbound calls that mostly did not connect. Qualification was the bottleneck, not closing.

Approach

A production voice agent on Twilio, Retell AI and OpenAI. Real time conversation handling, intent based routing, transcript capture, and CRM write back on every call outcome. Owned from architecture through live deployment.

Result

First touch qualification runs without a person in the loop. Reps pick up warm conversations only.

Python · FastAPI · Twilio · Retell AI · OpenAI · PostgreSQL

Problem

Catalogue and stock maintained by hand across two storefronts. Drift caused oversells, and updates ate a full working day each week.

Approach

A sync engine over both REST APIs with webhook driven deltas, queued reconciliation batches, idempotent writes and a replayable failure log. I led the team building it, then led an 80% migration from Laravel to Python.

Result

300,000+ listings held in sync. Manual updates down 80%, roughly 20 hours a week returned to the team.

Laravel · Python · MySQL · Shopify API · eBay API · Webhooks · AWSRead the full case study →

Problem

A healthcare platform needed infrastructure that could hold sensitive data without a later security review turning into a rewrite.

Approach

Built and deployed on AWS ECS with RDS PostgreSQL and ElastiCache Redis, inside a private VPC reached through a bastion host, with every credential held in Secrets Manager rather than the repository.

Result

A live platform on infrastructure designed for the compliance conversation up front instead of retrofitted for it afterwards.

Python · React / Next.js · PostgreSQL · AWS ECS · RDS · ElastiCache · Secrets Manager

Problem

Every order re-keyed into the courier portal by hand. Address typos became returned parcels, refunds and support time.

Approach

An event driven pipeline running order webhook, address validation, Royal Mail label generation and tracking write back, with retries and a dead letter queue.

Result

40+ orders a day dispatched with no human in the loop. Ops now touch only the exception queue.

Laravel · Queue workers · Royal Mail API · MySQL

Problem

Investor positions and hardware utilisation tracked in spreadsheets. Reporting was monthly, manual and already stale on arrival.

Approach

A Python service aggregating utilisation and financial events into Postgres, behind a Next.js portal with per investor scoping and live dashboards.

Result

The monthly spreadsheet cycle replaced by self serve reporting investors read themselves.

Python · Next.js · React · PostgreSQL

All case files →

Measured outcomes

Taken from delivered systems, not estimates

300K

Products in sync

40/d

Orders, hands off

80%

Manual work removed

Years in production

Live and public

MultSaver

The only thing on this site you can click and use right now. A free media downloader I run myself, with no signup and no upsell. It exists because the alternatives break constantly, and keeping it alive is a standing test of the reliability work I sell.

Open multsaver.com ↗

Platforms

Instagram, TikTok, Facebook, X

Cost to user

Free, no account

Stack

FastAPI, Redis, FFmpeg

Uptime work

Extractors isolated per platform

Each platform extractor is isolated, so one breaking never takes the service down. It is the same pattern I use in client integrations.

Who this
is for

Teams shipping AI features

A prototype that works in a notebook and now has to survive real users, real latency and real error handling.

Online retailers

Selling on more than one channel and reconciling stock by hand. Shopify, eBay, Amazon, WooCommerce.

Funded startups

Frontend covered, no backend hire yet, and an integration or pipeline blocking the roadmap.

Ops heavy SMBs

A team of two to twenty where somebody's whole morning is copy and paste between systems.

Moeez Rehman, Full Stack AI Engineer, photographed outdoors in Lahore — Moeez Rehman · Lahore, Pakistan

Who you would actually be working with

No account manager,
no handover.

You talk to me, and I write the code. There is nobody in between translating your problem into a ticket, and nobody junior quietly inheriting the work once the contract is signed.

Five years of it, including a stretch leading the developers who built a sync engine holding 300,000 listings in agreement. Most of that time has been spent inside other people’s systems, which is a good school for writing things the next person can actually maintain.

More about how I work →

How I work

Four steps, no surprises

Step 01

Map the process

A call where you walk me through the manual work. I write down each step and what it costs you per week.

Step 02

Scope in writing

What gets built, what does not, the price and the date. You approve it before any code exists.

Step 03

Build in the open

Weekly demo on a staging environment you can click through. No month of silence ending in a surprise.

Step 04

Hand over properly

Deployed, documented, alerting wired up, plus 30 days of support so it survives contact with reality.

Engagements

Full detail →

A / AI build

A voice or LLM agent, or an MCP server exposing your systems to one. Taken from prototype to something that survives production.

4 to 8 weeksQuoted after scoping

B / Integration build

Two or more systems that need to talk. Fixed scope, fixed price, documentation at handover.

3 to 6 weeksQuoted after scoping

C / Retainer

Standing capacity for teams without a backend or AI hire, and someone on call when things break.

Monthly, rollingQuoted after scoping

In their words

Client reviews, quoted as written

5.0

“Great experience working with Moeez on our voice AI agent feature. Strong Python's, FastAPI and backend skills, clean integration, and solid communication throughout the project.”

AI Engineer Needed for Voice AI Agent Feature (Python/FastAPI)

Upwork client · Aug 2026

AI Agent Development · Generative AI · AI Chatbot · Reliable

5.0

“Moeez was excellent to work with. He quickly identified the bug in our FastAPI backend and fixed it efficiently without breaking any existing functionality. Great communication throughout.”

Need Help Fixing a Small Bug in My Python Website (FastAPI)

Upwork client · Jul 2026

FastAPI · Committed to Quality · Solution Oriented · Clear Communicator

Teams I have
built for

Recent engagements, one of them as team lead

Jun 2025 - Aug 2026

Codiux

Full Stack Engineer

Lahore

Mar 2024 - Jun 2025

Maxenius Solutions

Backend Engineer · team lead

Faisalabad

May 2023 - Feb 2024

XLogic Solution

Backend Engineer

Faisalabad

References available on request.

Common
questions

Anything not covered here, ask on the call. It costs nothing.

Yes, openly. Claude Code, Codex and Cursor are part of my daily setup, and they make the mechanical work faster: boilerplate, refactors, test scaffolding, reading an unfamiliar codebase. What they do not do is decide what to build, judge whether an approach survives production load, or carry the responsibility when it does not. Everything that ships is code I have read, understood and can defend line by line. You are paying for that judgement, and these tools free up more of my time to spend on it.

Next step

LET’S BUILD
SOMETHING DULL.

The best engineering is boring. It runs, nobody thinks about it, and it pays for itself in returned hours. Tell me what you keep doing by hand.

Book 20 minutes →Email instead ↗

SYSTEMSTHAT RUNTHEMSELVES.

The toolkit

AI engineering

AI assisted development

Languages & frameworks

Backend & APIs

Databases

Cloud & DevOps

What I actually do

AI agents in production

MCP servers & agent tooling

API design & integration

Automation workflows

Full stack product builds

Cloud deployment & hardening

Selected work

Measured outcomes

MultSaver

Who thisis for

Teams shipping AI features

Online retailers

Funded startups

Ops heavy SMBs

No account manager,no handover.

How I work

Map the process

Scope in writing

Build in the open

Hand over properly

Engagements

In their words

Teams I havebuilt for

Codiux

Maxenius Solutions

XLogic Solution

Commonquestions

LET’S BUILDSOMETHING DULL.

Who this
is for

No account manager,
no handover.

Teams I have
built for

Common
questions

LET’S BUILD
SOMETHING DULL.