M M Asif Ferdous, Dhaka, Bangladesh
AI that answers your customers and follows up every lead.
Chatbots that reply day and night, automations that stop enquiries slipping through, dashboards that make your numbers clear, and the websites and apps they live in. I build them for clinics, restaurants, hotels and data-driven companies, and white-label for agencies.
- Experience
- 7 years in production
- Research
- 2 papers on AI reliability
- Works with
- USA, UK, Australia, New Zealand
- Work with me
- Directly, or white-label via agencies
Two ways to work with me
For businesses
I build it, set it up with your team and look after it afterwards. You deal with me directly.
- Dental and medical clinics
- Restaurants and cafes
- Hotels and guest houses
- Data-driven companies
- Law firms
- Shops and online stores
- A chatbot that answers patients, diners or guests at any hour
- Every enquiry and booking request replied to, saved and followed up
- One dashboard for bookings, sales and marketing
- A new website or booking app, with the chatbot built in
For agencies
Web and marketing agencies whose clients want AI. I build it white-label under your brand, never contact your clients, and you set your own margin.
- Chatbots, automations and reports, delivered as your work
- Partner rates and a small paid pilot first
- NDA on request
One enquiry, handled in 38 seconds
Here is a dental clinic after I set up a website assistant and a lead automation. The patient gets a real answer straight away, the enquiry lands in the CRM already sorted, and the front desk is told. Nobody copies and pastes anything.
The same setup works for restaurants taking table bookings, hotels answering guests, and any business that lives on enquiries.
Brightside Dental
Website enquiries, example client
Under your brand
New enquiry on the website
"Do you do Invisalign consultations on Saturdays?"
AI assistant replied
Answered from the clinic's own FAQ and offered two Saturday slots.
Lead sorted and saved
Tagged "Invisalign, ready to book" and added to HubSpot.
Front desk alerted
Message posted to the #new-patients channel in Slack.
What I build
The four things businesses ask for most. Agencies can resell any of them under their own name.
AI website assistant
Clinics, restaurants, hotels, law firms, and any business that answers the same questions all day.
A chatbot trained on your own website, FAQs and documents. It answers visitors accurately at any hour, books or captures the lead, and sends it where your team will see it.
- Answers only from your own content, so it does not make things up
- Lead capture to email, Google Sheets or the CRM
- Plain-English guide for your team, or under your agency name
- Price
- Build from $400
- Care from $60 a month per bot
Lead‑to‑CRM automation
Clinics, hotels, restaurants and service businesses that lose enquiries between the website, ads, WhatsApp and the inbox.
Every new enquiry gets an instant reply, is sorted by what the person wants, lands in the CRM and alerts the right person, without anyone copying and pasting.
- Website forms, Facebook lead ads, WhatsApp and email
- HubSpot, Pipedrive, Google Sheets and most other CRMs
- Alerts by email, Slack or SMS
- Price
- Build from $300
- Care from $40 a month
Data dashboards and visualisation
Data-driven companies, hotels and restaurant groups tracking bookings and sales, and agencies that build reports by hand.
Analytics, ads, sales and CRM numbers pulled into one dashboard that updates itself, with clear charts and a short written summary of what changed and why it matters.
- Bookings, sales, GA4, Google Ads, Meta Ads and CRM data in one place
- Interactive charts for websites and internal tools, like the one on this page
- Data cleaning and analysis in SQL and Python
- Price
- Build from $400
- Care from $50 a month
Web and mobile development
Clinics, restaurants and hotels that need a new website or app, and agencies with more work than their team can take on.
Websites for clinics and restaurants, booking systems, customer apps and admin panels, built properly and fast on any phone. Take a fixed-price project, or a block of hours every month.
- Business websites, web apps and Android apps
- React, Angular, Node.js and PHP
- Databases, logins, integrations and APIs
- A written log of hours and work each month
- Price
- Projects quoted at a fixed price
- Or 20 hours a month from $600
Starting prices in US dollars. Every job gets a fixed quote first, and agencies reselling my work get partner rates.
Data you can read at a glance
This chart uses real data from my own research: how well three AI setups spot their own mistakes on damaged photos. Switch between what a model knows and what it says, and watch the gap open up.
The same care goes into business dashboards: clean data, the right chart for the question, and nothing you have to decode.
- 2B, full precision
- 2B, 4-bit
- 7B, 4-bit
Score for spotting its own mistakes, from 0.5 (guessing) to 1.0 (always knows). Hover a dot for details.
How working together starts
- 1
A small paid first project
Pick one problem, such as missed enquiries or slow replies. I quote a fixed price, usually $300 to $600, and deliver in 7 to 10 days so you can judge the work properly.
- 2
Built around how you work
I set it up with your team and your tools, in your branding. For agencies, everything carries your name and your client only ever deals with you.
- 3
Ongoing care
If it works, I look after what I built for a small monthly fee, keep it up to date, and take on the next job when you need it.
What you can count on
- Agencies keep their clients
- For white-label work I never contact your clients unless you ask me to join a call, and I am happy to sign your NDA.
- Written scope for every job
- What is included, what is not, and the price. Changes are quoted before I do them.
- Simple payment terms
- Builds are 50% upfront and 50% on delivery. Monthly care is billed in advance. Invoices go out through Payoneer or Deel.
- Working hours that overlap
- I work from Dhaka (UTC+6) with flexible hours: UK mornings, Australian and New Zealand afternoons, and US mornings, which fall in my evening.
Selected work
AI systems, web platforms and mobile apps I have built and shipped.
Personalised cardiovascular risk model
Manuscript in preparationA two-stage machine learning system that screens for peripheral vascular disease, then estimates how much it raises each individual patient's cardiovascular risk.
Python, scikit-learn, XGBoost, Flask
LegalMind AI
Open sourceA question-answering system over legal documents. Upload PDFs, ask in plain English, and get answers with citations to the exact passages they came from.
Python, LangChain, FAISS, HuggingFace embeddings
Takamol exam-prep platform
LiveA bilingual Bengali and English practice platform that helps job candidates prepare for Saudi Arabia's Takamol skills exams.
React, mammoth.js, Firebase Hosting
FCI Communication App
ArchivedA pair of Android apps for my university faculty, connecting students and staff through direct chat, announcements and surveys, with instant push notifications.
Android, Java, Firebase Realtime Database, Cloud Functions
TradeFlow import tracker
LiveAn installable web app for tracking air-conditioning and refrigeration parts imported from China, built for an import business I run.
React, PWA, Firebase
Job search AI agent
Open sourceAn assistant that collects job listings, reads your CV, ranks the jobs that fit you best and tracks every application.
Python, Streamlit, scikit-learn, pdfminer
Demos
Working examples you can try yourself before deciding anything.
- Being built
AI website assistant
A chatbot trained on a real small-business website. Ask it anything a customer would, and watch it answer from the site's own content and capture the lead.
- Being built
Business reporting dashboard
The monthly report someone usually builds by hand, generated automatically from bookings, analytics and ads data, with a short written summary of what changed.
Research
I study when AI models can be trusted to answer and when they should hand over to a person. The same question sits under every chatbot I build.
Read about the papersarXiv preprint 2607.22034, July 2026
Small Vision-Language Models Know When They Are Wrong But Cannot Say So: A Two-Model Study of Stated versus Internal Confidence Under Realistic Image Degradation
Small vision-language models carry useful knowledge of their own mistakes in their token probabilities, even when the confidence they state in words is almost useless.
arXiv preprint 2607.24440, July 2026
Bigger or Cheaper? Scale and Quantization Effects on Uncertainty Signals in Vision-Language Models Under Image Degradation
On a fixed memory budget, a larger model compressed to 4-bit beats a smaller full-precision model on both accuracy and knowing when it is wrong.
Hi, I'm Asif
I spent more than six years as a software engineer at MobileOne Technology in Malaysia, building and running Angular, Node.js and PHP systems in production. In my own time I research the reliability of AI models, and I've published two papers on it.
You get both sides from me: web engineering that holds up in production, and AI used only where it genuinely saves you time.
More about meTell me what is slowing you down
Describe your business, or your client's, and the problem you want solved. I'll reply within one working day with how I would approach it and a fixed price to start.