ML Pipelines
End-to-end pipelines that turn raw, messy business data into scored, decision-ready output — ingestion, feature engineering, model, and reporting, built to run on their own.
AI systems, agents & automation
Intelligent systems that do the work your business doesn’t have time for.
I design and build AI agents, ML pipelines, and automation that turn repetitive operations into reliable, measurable outcomes — engineered end to end.
Most teams are buried in work their software was supposed to remove. I build the systems that actually remove it — and hand them over running.
Book a callWhat I build
AI Agents & Orchestration
Agentic workflows (LangGraph, n8n) that capture, decide, and act — replacing manual steps with systems that qualify, route, and respond automatically.
Data Analytics & Power BI
Dashboards and reporting layers that make the output legible to the people who act on it — KPI cards, filters, ranked tables.
Work — Systems I've designed and built
A selection of the agents, pipelines, and ML systems I've built end to end — two of them running live and available to walk through on a call.
Real Estate · Machine Learning
An ML pipeline that scores which property owners are likely to sell
Ingests 109,000+ county property records, engineers ML features, and scores every residential property 0–100 for seller intent using weak supervision and gradient boosting — delivered as a ranked weekly lead list in Power BI.
- ML Pipelines
- Data & BI
- Python
109K+properties scored 0–100 for seller intentRead case study
E-commerce · AutomationAn order pipeline that surfaces the margin data no-code tools can't
Pulls raw eBay Fulfillment API data into a structured Google Sheet — net profit per order, ad fees, buyer repeat status, fulfillment deadlines, refund flags — with custom business logic and one-time ownership instead of a monthly subscription.
- Automation
- API Integration
- Python
Full margin visibilityper-order data Zapier can't reachRead case study
Real Estate · AI AgentsAn agent that scores and routes inbound leads in seconds
A 7-stage workflow that captures a lead, validates it, scores it with an LLM (returning a score, category, and reason), routes it hot/warm/cold, stores it, and fires an instant alert on hot leads. Built on n8n + Postgres, verified end to end.
- AI Agents
- Automation
- LLM
92/100a live lead scored and alerted in secondsRead case study
What's been shipped
- properties scored 0–100 for seller intent
- 109K+properties scored 0–100 for seller intent
- a live lead scored, routed & alerted in seconds
- 92/100a live lead scored, routed & alerted in seconds
- systems running live
- 2systems running live
Available to walk through on a call — not slides, the real systems.
How I work
- 01 — Discovery
Map the real bottleneck and the outcome that matters, before writing a line of code.
- 02 — Architecture
Design the system end to end — data flow, models, agents, and where humans stay in the loop.
- 03 — Build
Implement and verify against that outcome, not a demo — tested on real data.
- 04 — Handoff
You own it, with documentation and monitoring so it keeps running without me.
About
Haqbil is the studio of Husnain Pasha — an AI systems builder who designs and ships vertical AI systems end to end. Not templated automations or demos: production systems sold on the business outcome they deliver. Based in Pakistan, building for clients worldwide.
If something in your operation should already be automated, let's look at it together.
Book a call