AI agents & workflow automation
Agents that take real actions: qualify leads, send follow-ups, update your CRM, review ad accounts. With approval steps where a wrong action is expensive.
AI systems that take repetitive work off your team: reading documents, qualifying leads, answering questions and reporting. I test them on real examples, log what they do, and add a human approval step wherever a mistake would cost money.
Each case study shows the problem, what I built, how it works and the proof. Open any card for the full breakdown.
Pick the problem, not the tool. I choose Make.com, n8n or Python based on what you already use and what will be easiest to maintain.
Agents that take real actions: qualify leads, send follow-ups, update your CRM, review ad accounts. With approval steps where a wrong action is expensive.
Answers from your PDFs, reports or pages with the source shown, and an honest “I don’t have that information” when the answer is not there.
Invoices, remittances, shipping papers and tenders turned into clean data, checked against your rules and sent to the right system.
Voice agents that call and qualify leads, and Telegram or WhatsApp bots that read photos and PDFs and reply in the chat.
Pull your CRM, ad platforms and sheets into one place that refreshes itself, so reporting stops being a weekly copy-paste job.
When a workflow tool is not enough: Next.js frontends, FastAPI backends, Supabase with row-level security, live data over WebSockets.
You see it working early, and you are never locked in.
We walk through the process as it runs today: what comes in, who touches it, where it ends up, and what a mistake costs.
I build a small working version on sample data or yours, so you can see it run before committing to the full build.
How I start every projectValidation, logs, retries and human approval where it matters. Tested against real examples, including the messy ones.
Built in your accounts where possible. You get documentation and a recorded walkthrough, and you own everything.
I’m a Computer Engineer from Larkana, Pakistan. I build AI agents, RAG chatbots and automations with Claude, GPT, Gemini, Make.com, n8n and Python. My work sits where AI meets real business processes: documents, leads, outreach, reporting and support.
I care about systems that keep working after the demo. That means structured outputs, logs, tests against real examples, and a human approval step wherever a wrong action costs money. When something should be calculated, like a price level or a total, I calculate it in code and let the model explain it.
I’m doing an MS in Computer Science at The University of Larkano and hold the Google Cybersecurity Professional Certificate, so security is part of how I build from the start.
Outside of code I’ve spent years in community healthcare and relief volunteering. It taught me to explain complicated things simply, which is most of what a good handover is.
Building AI agents, RAG chatbots and Make.com / n8n automations for clients and as reusable products. Client work includes a Google Ads approval workflow, a six-agent LinkedIn assistant, an AI newsletter pipeline and a market briefing bot. I start with a working demo, then build the full system with tests, logs and a proper handover.
Supported telecommunication systems and network operations inside a large corporate environment, including operational logs, system documentation and cross-department coordination.
Built custom Google Sheets automations in JavaScript to streamline data and reporting workflows. This is where my automation work started.
Coordinated healthcare access for families and supported them through the system.
Field logistics and supply management for relief work.
Organised university donation drives and managed the donor database.
The University of Larkano · 2nd semester · Advanced Database Systems, Advanced Theory of Computation
Balochistan University of Engineering & Technology (BUET), Khuzdar · CGPA 3.14 / 4.0
University of Balochistan, Quetta
Govt. Boys Degree College, Larkana
Govt. Boys MPL H.S. School, Larkana
Tell me what’s eating your team’s time. I’ll tell you honestly whether an AI workflow can fix it and what it would take.