Projects and case studies
Things I have built, backed and turned around.
Mission-critical platforms I architected and delivered, companies I founded, investments that reached an exit, and the work behind this publication.
Case studies
Systems that millions of people relied on
Architecture and delivery of platforms where downtime, wrong numbers or slow decisions have real consequences. Clients are not named; scope and impact are.
Taxi corporations in Central Europe
The platform that ran taxi companies end to end: every ride from the order to the invoice, orchestrated across passengers, dispatchers and drivers.
- Orchestration of millions of rides a month
- Mobile apps for passengers and apps for hotels ordering rides
- Dispatcher application and driver apps for accepting orders
- Payment cards, vouchers and customer loyalty cards
- Customer billing and invoicing
From the ride order to the invoice
- Passengers order a ride in the mobile app.
- Hotels order rides for their guests from an app of their own.
- Dispatchers take orders by phone in the dispatcher application.
- Every order, whichever channel it came from, lands in one orchestration engine that handles millions of rides a month.
- The engine sends the order to a driver, who accepts it in the driver app.
- The ride is paid by payment card, voucher or loyalty card, and the payment is recorded against the ride.
- Billing turns rides and payments into customer invoices, so operations and finance work from the same record.
An AI voice agent that takes taxi orders by phone
A voice agent that answers a taxi company's phone line, talks with passengers in natural language and books the ride straight into the dispatch system. Built for taxi companies since 2019.
- Natural conversation on the company's existing number, with no "press 1" menus, in Polish and English
- Caller recognition, pickup address, price and arrival time from live data
- Bookings, rides for later, "where is my taxi", changes and cancellations
- Integration with taxi dispatch systems and hand-off to a human dispatcher
From a phone call to a booked ride
- A passenger calls the taxi company's usual number. No new number, no app and no "press 1" menu.
- The voice agent understands natural speech, lets the caller correct themselves, asks when something is missing and switches between Polish and English.
- It builds the order from context: it recognises the caller by number and ride history, finds the pickup address, quotes the price from the company's price list and estimates arrival from free cars nearby and traffic.
- Before booking, it reads the order back and the caller confirms it.
- The ride is booked in the dispatch system. The same agent checks where a taxi is, changes or cancels a ride, or books one for later.
- When the caller asks for a person, or the case needs a decision, the call goes to a human dispatcher. The agent can answer every call, day and night, or only those the dispatchers cannot pick up.
Train traffic management
A system that managed more than 1,000 trains at a time, together with the infrastructure they run on.
- Infrastructure model: tracks, level crossings and switches
- Live traffic management and analysis of risks and delays
- Settlement between rail operators, including penalties for disruptions and delays
- Billing for track time and consumed resources
From the track model to the invoice
- Everything starts with a precise model of the infrastructure: tracks, level crossings and switches, with their state.
- Traffic management runs on top of it, for more than a thousand trains at the same time.
- Every movement feeds the analysis of risks and delays, so operators see problems before they cascade.
- The same data settles accounts between rail operators, including who pays for disruptions and delays.
- Finally, track time and consumed resources are billed. One consistent record serves operations and finance.
Leasing sales for a major car brand
The system behind leasing sales for one of the largest automotive brands.
- Product management and offer management
- Instalment and residual value calculation
- Decision management for leasing applications
- Dealer-facing sales workflow from quote to signed contract
From the offer to the signed contract
- Leasing products and offers are managed in one place: what can be leased, and on what terms.
- The dealer builds a quote for the customer inside the sales workflow.
- Instalments and the residual value are calculated for the chosen car and terms.
- Decision management evaluates the leasing application.
- An approved deal ends with a signed contract: quote to signature in one workflow.
Connected fleet, driver scoring and risk
A connected-fleet platform for several large insurance companies and banks, with tens of thousands of vehicles under management: from the tracker in the car to risk, cost and portfolio reports.
- Telematics from GPS trackers, plug-in OBD devices and manufacturer data
- Live location, route history, geofences and alerts
- Driving-style and eco-driving analysis, detection of dangerous situations, driver risk scoring
- Gamification that rewards safer driving
- Fleet operations: service and inspections, documents and deadlines, costs, car pooling and mileage logs
- Portfolio dashboards, reporting and an open API for insurers and lenders
From the car to risk, rewards and reports
- Vehicles report through GPS trackers, plug-in OBD devices or directly from the manufacturer's connected-car data.
- The telematics core turns the raw signal into trips and driving events, keeps live location and route history, and raises alerts, for example when a car leaves a geofence.
- Driving style, eco-driving and dangerous situations are analysed and turned into a risk score for each driver.
- Drivers see their score through gamification that rewards safer driving.
- Fleet managers run operations on the same data: service and inspections, documents and deadlines, costs, car pooling and mileage logs.
- Insurers and lenders get the scores and costs rolled up in portfolio dashboards and reports, or through an open API into their own systems.
Simplified architectures. Client and system names are confidential.
Built
Founded and led
PlaceChallenge
A platform for location-based games. I founded it, built the product and sold the company.
Outcome: sold.
CloudExpert
My base for cloud, architecture and software work with companies since 2011, and the practice behind my interim and advisory roles.
Focus: architecture, cloud, product engineering.
DevInside
This publication: long-form engineering articles with verified code, a custom Ghost theme, and an inline motion engine with an automated visual validator.
Read: the articles.
Backed
Investments that reached an exit
A selection from the portfolio. The full list, with the outcome of every investment, is on the investing page.
Preply
Backed at seed; the stake was sold in 2022 at 14x cash-on-cash, as reported by the fund's partner.
Allani
Acquired by Domodi, part of the Wirtualna Polska group; Forbes described the return as roughly 6x.
ZenCard
Card-linked loyalty platform acquired in full by PKO Bank Polski.
XTRF
Translation management platform; the fund's stake was bought out in 2020, before K1 acquired the company.
Perfect Dashboard
Automated website updates; technology and team acquired by WP Engine.
Skilltelligence
Data collection and annotation for AI; acquired by TransPerfect for its DataForce business.
Turned around
Interim and board engagements
Restructuring products, departments and IT, building new products and serving on supervisory and management boards.
These engagements are covered by confidentiality, so they are not listed here by name. If you are considering me for a similar role, I am happy to describe comparable situations and put you in touch with references.
Community
Microsoft Regional Director, former MVP
Since 2014 I have been part of the Microsoft Regional Director programme, an invitation-only group through which Microsoft works with independent technology leaders around the world.
Before that I was a Microsoft Most Valuable Professional (MVP) from 2008. Speaking, writing and working with developer communities have been part of the job from the beginning. DevInside is the current form of that work.
2008 to 2014Want details on a specific project?
Ask about a company, a role or a type of engagement and I will tell you what I can.