Skip to content
Data engineerBased in Belgium

I'm a data engineer who builds for people who don't write code. Spreadsheets, scripts and repeated steps become dashboards, automations and simple apps, designed around how the team already works.

Hello, that's me ↙
Danil Zhuravlov
Curious about how things work. And how they could work better.
What I've built, and why.Take a look below ↓

01 / Selected work

Built from
a real reason.

Different projects. The same starting point: something could be easier, clearer, or more useful.

01 / Production overviewProduction pipeline / Internal tooling

The whole process.
At a glance.

To run the next step of our production pipeline, you had to know the entire process. Which steps were done? Which were still running? What did we already have, and what could we start based on it? The answers were spread across logs, created files and people's memory.

I built a dashboard that shows every step of the pipeline and its status: not ready yet, ready to start, running, or finished.

Now nobody has to keep the whole process in their head. The dashboard shows what's happening and explains which actions you can take next, and which you can't yet.

Result

From knowing the whole process to seeing the next step.

02 / Silent failuresData pipelines / Reliability

Failures that show up.
Before they cost a day.

A pipeline step could fail without an error, leaving an empty output file behind. Nobody noticed until the next step needed that data. Usually a day later, sometimes two or three.

I added checks that catch the failure and its reason, and write it to the status spreadsheet the team already uses.

Now the problem is visible as soon as someone opens the spreadsheet. In practice that's half a day, a day at most. It isn't an alert yet, so the next step is making failures come to people instead of waiting to be found.

Result

From 1 to 3 days to under a day.

03 / Renaming automationWorkflow automation / Internal tooling

Two hours of renaming.
Five minutes of script.

Every Monday, a colleague spent about two hours taking project files and renaming them by the same systematic rules. Same structure every week, different input for each project.

I turned those rules into a small Python script. Now it's one run of around five minutes.

It's a modest piece of software with a clear purpose: take the mechanical steps out of a routine, and give that time back to work that needs a person's judgment.

Result

About 2 hours a week down to 5 minutes. Roughly 90 hours a year.*

* Estimate, based on about 46 working weeks a year.

04 / CRMInternal tool / Adoption by choice

Built around people.
And their process.

I noticed sales colleagues struggling with their existing tools and Excel-based processes. Nobody had assigned me a CRM. I started building one.

I adapted the application to the way they already worked. They could keep using Excel, try the CRM, or combine both. Moving everything at once was never the condition for getting something useful.

The application exists; becoming the preferred choice is still the goal. I want real use and feedback to guide what comes next, until the value of using it is a reason to switch.

Keep ExcelUse bothTry the CRM
A design decision: give people time and choice.

Side projects

Built for myself.
Tested on myself.

Personal projects where I try ideas first, with me as the only user.

05 / TaskTreePersonal project / Desktop application

My work.
In context.

The Microsoft 365 tools I had did not fit how I wanted to track priorities, project stages and completed work. So I built TaskTree around that view of my work.

Projects break down into nested tasks, with statuses and priorities that keep both the next action and the larger picture visible. Completed work stays part of the story.

I can use the interface or operate the application through conversation: ask about tasks, change a priority, or update a status. The assistant uses the same application API, so it works on the same task data. Python, Flask and SQLite keep the application and its data together on the desktop.

06 / JinPersonal project / Telegram

A familiar place
to pick things up.

I wanted help with everyday responsibilities without losing sight of my longer-term ambitions. I also did not want another application to maintain.

So I chose a familiar Telegram conversation. Jin has a name, a consistent personality and proactive responsibilities. Those are interaction decisions: how he remembers context, when he checks in, and when he should leave me space.

The current implementation includes tasks and reminders, saved project context and drafts, and bounded proactive suggestions with quiet hours and pause controls. It is a private project for my own use. How helpful that interaction feels over time is something I am still developing through use and feedback.

A Telegram conversation with Jin about a train journey and weekend plans. A personal name is redacted.
Jin in use: checking a journey and recalling weekend plans. Personal name redacted.

02 / How I work

Software should
earn its place
in your workflow.

I start by understanding how people already work and where things become difficult. Then I build a useful first improvement, introduce it gradually, and refine it with the people using it. Straightforward interfaces and automation help reduce effort without forcing everyone to change their process overnight.

  1. 01

    Notice the friction.

    I look for repeated effort and missed opportunities, including things nobody has put in a specification.

  2. 02

    Start with something useful.

    I build a manageable improvement around the team's existing process, with an interface people can use and fewer manual steps.

  3. 03

    Let real use and feedback shape what comes next.

    I listen, adjust and add incrementally. People need time and choice; the tool should give them a reason to use it.

03 / A little about me

Built from
the outside in.

Ukrainian. Based in Belgium.

I'm a data engineer. Since March 2025, I've worked in geospatial production for aerial survey.

My work includes automation, applied machine learning and quality control before deliverables reach clients. Building for colleagues who do not write code has made usable interfaces a practical priority.

I'm Ukrainian and now based in Belgium. The war changed my life profoundly. Being Ukrainian remains an important part of who I am, and those experiences have strengthened my determination to grow and contribute something meaningful through my work.

I came into engineering from outside it. I remember what it feels like not to know how the software works, and I try to keep that perspective when I build.

View my CV

04 / Let's talk

Have a workflow that feels harder than it should be? I'm always happy to talk about it.

Get in touch