Taking new projects · remote, EU & Asia hours

More attention.
Less busywork.

// Fahad Asad · AI automation

I help creators and small businesses grow on two fronts: content that wins the click and holds attention, and AI tools, dashboards and APIs that take hours of manual work off your team and show you where revenue is leaking.

~/fahad/engagement.ts
const project = {
  you_bring: "a goal and a deadline",
  i_deliver: ["working product", "source code",
              "handover docs"],
  measured_by: ["click-through", "watch time",
                "hours saved", "revenue kept"],
  updates:   "weekly, in plain English",
  background: "Digital Economics & Business",
};

// $ npm run start-project
✔ plan ready, build started
76%of churners flagged in advance by Churn Radar
13k/srequests served by one Snaplink container
43%less forecast error than the rule of thumb (Sales Forecaster)
8products shipped, all with source code
Services

What I can do for your business.

Four ways I usually help. Each one starts from the number you want to move, not from the tool.

AI tools & automation

Goal: hours back every week.

Custom assistants built on the Claude API that read your documents, screen CVs, draft replies or answer staff questions, with sources your team can check.

  • Document Q&A with citations
  • CV and application screening
  • Internal tools wired into your data
Good fit if your team copies, pastes or searches the same information every day.

Data & revenue analytics

Goal: keep more of the revenue you already have.

Forecasts, customer segments and churn scores that turn your sales and customer data into decisions: what to stock, who to contact and what to offer.

  • Sales and demand forecasting
  • Customer segmentation and lifetime value
  • Churn and revenue-at-risk models
  • Dashboards: CSV in, action list out
Good fit if you sell to repeat customers or carry stock.

Thumbnails & packaging

Goal: higher click-through rate.

Thumbnails and titles designed as a pair, built around one clear promise so the right viewer clicks in under a second.

  • Concept options per video
  • A/B test variants
  • Channel-wide visual system
Good fit if your videos are good but the views don't show it.

Scripts & editing

Goal: longer watch time.

Hooks, structure and pacing written to keep people watching, then edited so nothing drags. One person from idea to export.

  • Research-backed scripts
  • Hook rewrites for existing videos
  • Editing for YouTube and shorts
Good fit if viewers click, then leave in the first 30 seconds.
Case studies

Business problems, working solutions.

Each project starts with a cost a business actually has: empty shelves, wasted discounts, lost customers, staff hunting through documents. All source code is open, so you can check the work.

All repos ↗
Business analytics
↗
Demand forecasting · Inventory

Sales Forecaster

ProblemRetailers reorder by rule of thumb, so shelves run empty in peak weeks and cash sits in the stockroom the rest of the time.
BuiltA forecast of the next 28 days for every store and category, a promotion what-if, and a reorder plan with safety stock you can download.
ResultWeekly forecast error cut from 21% to 12%, a 43% improvement over the rule of thumb, with almost no bias.
Weekly WAPE 12.3%6-fold backtestCalibrated 80% range

Measured on a rolling backtest over two years of realistic synthetic retail sales.

▶ Try the live demo
Pythonscikit-learnTime seriesStreamlitDocker
Sales Forecaster dashboard showing a 28-day revenue forecast with an 80% range
↗
Marketing analytics · CLV

Customer Segments

ProblemShops send everyone the same discount, wasting margin on loyal buyers and missing the valuable customers who are quietly leaving.
BuiltRFM segments checked against k-means clusters, a lifetime value and buy-again model, and a playbook with an exportable list for every segment.
ResultShows that 20% of customers bring in 69% of revenue, and flags 734 lapsing high-value customers worth about €87k every six months.
9 segmentsBuy-again AUC 0.79Out-of-time tested

Built on two years of realistic synthetic orders for 5,000 customers.

▶ Try the live demo
Pythonpandask-meansscikit-learnStreamlit
Customer Segments dashboard with a segment treemap and share of revenue by segment
Revenue & operations
↗ Retention analytics

Churn Radar

ProblemSubscription businesses lose customers quietly, and find out only when the revenue is gone.
BuiltA model that scores every customer's risk of cancelling, totals the monthly revenue at risk, and ranks the retention offers most likely to keep each one.
ResultFlags 76% of customers who go on to cancel, so a retention team knows exactly who to call first.
Recall 0.76ROC-AUC 0.83Revenue-at-risk view

Measured on a held-out test set of realistic synthetic subscription data.

Churn Radar dashboard showing churn rate, high-risk customers and monthly revenue at risk
▶ Try the live demo
Pythonscikit-learnpandasStreamlitDocker
↗ Internal tools

TaskHub API

ProblemTeams track work in spreadsheets with no access control or reporting.
ResultA secure backend for a team task tool, with logins, projects and progress stats, ready to deploy.
GET /projects/1/stats ● completion 50%, overdue 1
FastAPIPostgreSQLJWTDocker
↗
Marketing infrastructure

Snaplink

ProblemCampaign links need to survive traffic spikes and report clicks, without a big hosting bill.
BuiltA branded link shortener with click tracking, expiring links, rate limiting and live metrics, ready for Kubernetes.
ResultAbout 13,000 requests a second at 11 ms (p99) from a single small container.
GoRedisKubernetesPrometheus
Snaplink web UI showing a shortened link with 42 clicks
AI automation
↗
Knowledge assistant

AskMyDocs

ProblemStaff lose time digging through policies and contracts, and AI answers can't be trusted without a source.
BuiltUpload PDFs and ask questions in plain English. Every answer links to the exact passage it came from.
ResultAnswers in seconds that anyone can verify in one click, with prompt caching keeping follow-up questions cheap.
Next.jsClaude APICitationsPrompt caching
AskMyDocs answering questions about a remote work policy with numbered citations
↗ Hiring

ResumeFit AI

ProblemReading every CV against a job post by hand is slow and inconsistent.
ResultA fit score out of 100, missing skills and ATS keywords in seconds, with the same criteria applied to every candidate.
ResumeFit AI showing a fit score of 72 with strengths and missing requirements
Next.jsTypeScriptClaude structured outputs
↗ Engineering reporting

GitHub Pulse

ResultManagers ask Claude for repo health and a weekly activity report instead of building dashboards.
> repo_health ✔ CI passing ● score 92/100
TypeScriptMCP
Latest on GitHub
Content · Thumbnails

Packaging that earns the click.

The thumbnail is the ad for your video. Every one here was made for a real upload. Click any one to see it full size.

View all ↗
Content · Scripts

Watch time is written first.

A strong hook and tight structure keep viewers past the first 30 seconds. Open a script to read a real excerpt.

View all ↗
Process

Simple, clear, no surprises.

Free intro call

20 minutes on your goal, your numbers and your deadline. No pitch.

Fixed plan & quote

A written scope, timeline and price before any work starts.

Build in the open

Weekly updates you can see and click, so feedback lands early.

Handover

You own everything: files, source code and a short guide to run it.

YouTubers & creatorsThumbnails, scripts, editing
StartupsMVPs, APIs, AI features
Small businessesAutomation, dashboards
AgenciesWhite-label design & dev
About

A developer who thinks in revenue.

I'm Fahad Asad. I study Digital Economics and Business at the Università Politecnica delle Marche in Ancona, with A Levels in Computer Science, Maths and Physics.

That mix shapes how I work. Before I design or build anything, I ask what it should change for the business: more clicks, more watch time, fewer hours lost, fewer customers leaving. Then I build the simplest thing that moves that number.

Clients get one person who can write the script, design the thumbnail, edit the video and build the tool behind it, without handoffs between four freelancers.

NOW

Freelance developer & content partner

AI tools, analytics, APIs, thumbnails, scripts and editing.

NOW

Digital Economics & Business

Università Politecnica delle Marche, Ancona

NOW

A-Level tutor

Cambridge and Edexcel Computer Science, Maths and Physics.

BEFORE

A Levels

Computer Science, Maths and Physics.

Do you work with clients outside Pakistan and Italy?

Yes. Everything runs remotely, and splitting time between Karachi and Ancona means overlap with both European and Asian working hours.

How do you price projects?

A fixed price per project or per batch of thumbnails or scripts, agreed in writing before work starts. Ongoing work can move to a monthly retainer.

Do I own the code and files?

Yes. You get the source code, design files and a short guide, with no lock-in.

Can you work with my existing team?

Yes. I can plug into your GitHub, Slack or Drive and work alongside your developers, editors or marketers.

Start a project

Tell me the goal. I'll send a plan.

Email me a few lines about what you want to change. I'll reply with questions or a short plan, and we can book a free 20-minute call.

GoalMore views, hours saved, fewer cancellations…
TimelineWhen you need it live
BudgetA rough range is enough
m.fahadasadshaikh@gmail.comWrite an email →