Services

What I can build for you

Six kinds of work, one engineer. Each section says who it's for, what you receive and where to see similar work. If you're not sure which fits, answer one question or just describe the problem.

What I do

Six kinds of work

Each one links to real work, an article and something you can try.

01

Data extraction from difficult and protected websites

You need records from a website, portal or set of PDFs, and the easy approaches have not worked. Maybe it pages through results one click at a time, builds everything with JavaScript, sits behind a login you own, or runs bot detection such as Akamai or DataDome that blocks scripts or quietly serves them the wrong page.

What you get

  • An access diagnosis before any build: what the site checks, where simple collectors fail, and the most reliable route that fits your use
  • A clean CSV, Excel, JSON or database export, with the fields you asked for
  • The script or service that produced it, so you can run it again
  • Checks that compare the output to the source, so gaps are visible
  • Resumable runs: an interruption does not mean starting over
02

Browser and workflow automation

Someone on your team spends hours a week in the same web portal, spreadsheet or inbox: logging in, filling forms, downloading reports, copying values between tools.

What you get

  • A bot that follows your real workflow, including logins, multi-step forms and downloads
  • A human hand-off wherever a person should decide (for example, confirming a payment)
  • Notifications by email, Slack or Telegram when something needs attention
  • Logs and screenshots so you can see what it did
03

Backends, APIs and custom tools

You need a backend for an app, an internal tool, an API around your data, or a service that does work in the background on a queue or schedule.

What you get

  • A documented REST API (FastAPI, Django REST Framework or Go)
  • Background jobs with retries, idempotency and clear failure states
  • PostgreSQL, MySQL, MongoDB or SQLite schemas that fit the data
  • Deployment on your server or cloud, with a short runbook
04

Integrations and data pipelines

Your systems do not talk to each other. Orders, leads or reports get re-typed between a store, a CRM, a courier, a spreadsheet and an inbox.

What you get

  • Integrations with third-party APIs and webhooks, including the awkward ones
  • Handling for the cases that break naive integrations: timeouts, duplicates, uncertain results
  • Scheduled pipelines that clean, deduplicate and deliver data where you need it
  • Google Sheets, Excel, email or database outputs
05

Monitoring and alerts

You want to watch competitors, marketplaces, product availability, new listings or public records, and be told only when something real changes.

What you get

  • Scheduled checks with history, so changes can be compared over time
  • Alerts by Telegram, email or Slack, with the evidence attached
  • Data-quality checks so a failed page load is not reported as a price change
  • A simple dashboard or spreadsheet view
06

Infrastructure, reliability and rescue

You have a scraper, bot or pipeline that breaks often, gets blocked, runs too slowly or costs too much to operate. Or you need the infrastructure underneath one.

What you get

  • A diagnosis in writing: what is failing, and why
  • Fixes for stability: retries, backoff, checkpoints, health checks
  • Proxy and network setup, including self-hosted options
  • Docker, Linux servers, Raspberry Pi, AWS Lambda or a VPS, whichever fits the budget
Specialty

Protected and difficult sources

Generating a scraper is easy now. Keeping one correct against an application that changes, checks sessions and runs bot detection is engineering work. This is how I approach it.

  1. Investigate first

    Map the real request flow, session and cookie behaviour, rendering and pagination. Identify what the site checks, whether that is Akamai, DataDome, a challenge page or its own rules, and where simple collectors go wrong.

  2. Agree the route

    Choose the most reliable approach that fits your use and the source's terms: an official API or export where one exists, the application's own data requests, a real browser where needed. Trade-offs in speed, cost and upkeep are explained, not hidden.

  3. Build for wrong answers

    Every response is checked before it is trusted. Challenge pages, sign-in pages and partial records are detected, retried with care or flagged, and totals are reconciled against the source.

  4. Keep it working

    Monitoring that notices when the site changes, clear failure reports, and a runbook. Protected sources change; maintenance is planned, not a surprise.

What I won't do: use someone else's credentials, collect sensitive personal data without a lawful basis, or sell a general-purpose bypass tool. Ordinary scraping, automation and backend projects are just as welcome; this specialty is why the hard ones go well.

See the diagnostics demo →

Working together

Practical answers before you ask

How does it start?

You describe the problem. I reply within two working days, free, with whether it is doable, the risks and a sensible first step.

Do I need a paid sample?

Only when the source or system is genuinely uncertain. Then a small paid proof of concept on the real thing comes before the full build. Many projects go straight to a quote.

How is it priced?

Fixed price for well-defined work, with written acceptance checks. Hourly for open-ended or exploratory work. You will know which before anything starts.

Who owns the result?

You do: the code, the data and the documentation. Nothing is locked to my servers unless you ask me to host it.

What about maintenance?

Websites change and integrations drift. Ongoing maintenance is available, never required. Each handover includes a short runbook.

What won't you build?

Anything that needs someone else's stolen credentials, collects sensitive personal data without a lawful basis, or is designed to defraud. I will say so early if a request is in a grey area.

Next step

Not sure which of these fits?

Describe the work in plain words. I will tell you which approach fits, or honestly say if it is something I should not take on.