Start the change.
Now.

Trengo is a customer engagement platform that unifies conversations from every channel — email, WhatsApp, live chat, social — into one collaborative inbox, helping teams respond faster and automate support with AI.

As Head of Engineering, I lead the engineering organization: scaling the teams, the platform, and the delivery process to support Trengo’s growth.

zenloop, a German customer-experience startup, had built a SaaS product that rapidly gained traction — nearly 100 employees and almost 1,000 customers, funded through successive VC rounds. Just before their latest raise, the CTO and several core team members abruptly departed. I was brought on as the new CTO, tasked with scaling both the team and the product.

Two problems stood out. Team structure: technical decisions had been centralized at the CTO / senior-lead level, and that top-down, micromanaged culture had eroded autonomy. Developers owned features in silos; the nominal “squads” were really separate frontend and backend teams with little interplay.

Product architecture: a fast MVP had been overloaded into a monolith with convoluted code and a simplistic database — a single table had grown to billions of records, causing serious stability and performance issues. The team’s bandwidth was consumed by urgent bug fixes, leaving little room for new features.

I set two OKRs: cut idea-to-deployment for features to under two months, and reach 99.8% SLA availability (goal 99.99%). I ran extensive agile training and reorganized teams one at a time rather than all at once — each success made the next transition smoother. We moved from GitHub Issues to Jira and trained the product managers.

Despite the progress, several financial misjudgments by the CEO — against the wider management team’s advice — led zenloop to declare insolvency in early 2023. My role pivoted to managing the scale-down and preparing for due diligence and acquisition, which concluded in April 2023.

The Searchmetrics SEO Suite (Suite 7) offered an impressive array of tools, but like many specialized products it could overwhelm anyone not fluent in SEO. I introduced “Searchmetrics Actions” — a product that distilled Suite 7’s comprehensive data into an accessible, guided experience for fixing SEO issues on your domains.

With no pre-existing blueprint, the vision was clear: work independently at first, with the potential to integrate into Suite 7 later, and stay free of legacy constraints using state-of-the-art tech and the latest agile practices. In under two months we launched our first MVP and engaged beta customers to refine it.

Features were framed as automated workflows: a user could pick an objective like “boost backlinks,” enter a domain, and receive detailed recommendations, next steps, and an aggregated domain score — then implement and re-evaluate to gauge improvement. It resonated quickly, many “actions” followed, and it was folded into the Suite as a staple, renamed Searchmetrics Discover.

After the successful launch of Research Cloud, the owner of Searchmetrics proposed a product for web-page analysis — a domain crawler. The market had analogues, so we differentiated by excelling where others fell short: JavaScript crawling, a recommendation system, comparison against previous crawls, and competitor analysis.

I assembled a fresh team blending new talent with experienced hands, half remote and half on-site — a multifaceted setup that made the early stages challenging. We agreed to start with cutting-edge technology and a tight core-feature set to expedite planning and delivery.

We shipped the MVP swiftly as a Beta preview. Early feedback was stringent — customers compared us against established products and flagged feature gaps — but we held to the original vision of innovating rather than merely emulating. Through a year of beta and diligent feedback, we integrated the product into the Suite 7 Enterprise offering for our entire clientele.

In 2017 I joined Searchmetrics, then the premier SEO platform in the EU with a strong US presence. I quickly identified vulnerabilities: the platform ran on legacy code and an antiquated PHP version, making scalability hard and causing persistent performance issues. Development was split across four disconnected teams — Frontend, Backend, Data, Product — with suboptimal synergy.

The former owner, one of the world’s foremost SEO experts, wanted faster product enhancement. I proposed rolling out remarkable features while building a high-performance team to realize them — based in Croatia, my homeland. I secured an office and, tapping the local IT faculty (400 graduates a year) and my former professors, built a team of three fresh graduates and a seasoned developer in three months.

Our first project, Research Cloud, delivered detailed analyses of domains and keywords across 30+ countries over 15 years. Using a bottom-up approach, we upskilled the team while moving fast, and shipped the MVP in three months — instantly embraced by the market.

By year’s end the product had 30+ significant features, refined continuously by user feedback and usage monitoring, and it stayed the most popular module in the enterprise lineup for four years. Its success became a paradigm for the company: we established six full-stack teams, each owning a segment, laying the groundwork for “Suite 7.”

Croatian Telecom, the nation’s leading telecom and a Deutsche Telekom subsidiary, sold mostly through brick-and-mortar outlets and marketed via billboards and TV — sidelining the younger demographic that consumes the most mobile. Its web portal existed but wasn’t optimized for mobile.

I advocated for a stronger digital focus: not just sales or self-service, but tailored services for each customer via a mobile app. The proposal was approved. Rather than rely on the vendor-driven, hierarchical IT department, I recommended an agile approach with a compact internal-and-external team in my office, building for Android and iOS with a design adaptable across telecoms and backends — aimed at the broader Deutsche Telekom audience across the EU and US, and a precursor to a later SAFe implementation.

Within three months the app launched on both platforms with the most-wanted features. We iterated on user feedback; consistent 4-star-plus ratings on both stores proved the value of a nimble, flat-hierarchy team. After Croatia, it rolled out to Germany, Greece, and other EU markets, and today serves a user base exceeding 40 million.

Croatian Telecom struggled with the flow of goods between retail channels — Webshop, phone sales, brick-and-mortar, and wholesale warehouses. With no integration, deliveries dragged on: average delivery exceeded three weeks and NPS was negative.

I traced the root cause: only two people worked the warehouse, there was no integration between retail apps and the SAP warehouse system, and data moved via Excel over email. The channels didn’t communicate — the sales forces competed against each other.

Solving it took several parallel projects: integrating the SAP warehouse with retail apps and Webshop, plus educating sales teams to collaborate and training more warehouse staff. Average delivery time dropped from three weeks to three days, NPS rose 40%, and — most importantly — salespeople began working together, redirecting customers to another shop or home delivery when short on stock.

Croatian Telecom was an old, paper-dependent company — every contract change required a customer signature on paper. Becoming truly paperless meant many big steps; the first was cutting paper on the sales channels, its biggest producers.

I proposed digitizing all customer contracts and signing them on a tablet with a biometric signature. Because digital signatures and biometric data exchange were legal grey zones, I organized a tour of workshops with lawyers, judges, and forensic handwriting experts. Once they confirmed the biometric digital signature was valid, convincing the organization was easy.

End result: all contracts digitized and signable on a tablet, with a 20% rise in customer-satisfaction NPS.

A prominent sawmill and furniture company — three factories, 800+ employees — ran its entire business on paper forms. Management couldn’t track order status and suffered frequent production delays.

I assessed manufacturing, warehousing, logistics, and finance and recommended an ERP. With budget constraints ruling out SAP, I chose OpenERP (now Odoo) and selected a capable startup vendor, Info3, who were strong at software but new to ERP. Working closely, I led the implementation, collected specifications, and guided them in turning specs into solutions. Six months in, Info3 was proficient enough for me to hand over and move on.

On completion, revenue surged — customers loved the new online ordering system for large wood-material and furniture orders, and delivery promises were consistently met. Info3 grew from a handful of staff to more than 30 in under three years on the strength of the ERP expertise they gained.

PSA Financials financed automobile purchases through loans and leasing. Its European operations leaned heavily on paper, with limited digital support from the central banking system in France. The average wait for car financing reached nearly three weeks, breeding mistrust and pushing clients to competitors.

I was brought in to fix the paperwork overload and streamline processes between dealerships and the bank, leveraging the AbitECM system I’d helped build the previous year. We automated risk assessment and contracting and gave dealerships web access to start applications — so customers never had to visit a bank. Documents were scanned at the dealership, transmitted electronically, and processed by automation that extracted data for risk analysis and drafted contract proposals, cutting human error. Bank staff reviewed the final contract and risk analysis, then sent it to the customer to sign before car pickup.

The average waiting period dropped from three weeks to just 10 minutes, making PSA Financials a market leader in service.

Numip, a 150+ employee engineering company, services nuclear plants and builds automated manufacturing lines for pharmaceutical companies. Both sectors demand meticulous planning and documentation — a single plan could involve 50+ contributors. The goal was to speed up documentation without sacrificing quality.

Using the AbitECM system I’d helped develop the prior year, we built numerous automated processes tailored to Numip and — given the sensitivity of nuclear work — added high-security measures based on military-grade protocols.

The time to prepare standard projects fell by up to 40%, positioning Numip as a market leader able to service nuclear plants across Europe.

Abit Ltd built core banking systems, but its active products were legacy Informix 4GL “black screens.” Revenues were stable, yet as banking moved online there was real risk that banks would switch to more modern cores.

The company had started an internal project to build a new Java-based core, but without modern methods progress stalled and the project seemed interminable. Finishing my AbitECM work at the same company, I stepped in: I restructured development into three agile teams, each owning a major module, and secured a partner bank eager to collaborate as our primary customer — providing the essential business insights and requirements.

The project regained momentum and completed relatively quickly. It launched at the partner bank and was then adopted by several others, boosting Abit’s revenue and drawing interest from prominent European fintechs, leading to the company’s eventual acquisition.

AbitECM was the codename for a project that secured a grant from the EUREKA Network — the world’s largest public network for international R&D cooperation, spanning 45+ countries. Its aim: a system to automate business processes while managing diverse content types (documents, images) within those processes. Abit Ltd won the grant and recruited me to lead as lead developer, agile coach, and project manager.

The project completed on time and within budget. In its first year, nine clients — primarily banks, but also other businesses — adopted the system, and in subsequent years it significantly boosted company revenue until the eventual acquisition.

I launched Bedex-TIM to address two market gaps: genuinely effective smart systems, and a full service offering including installation and integration. Selling a complex product without accompanying services, I reasoned, would find limited traction.

As lead architect and developer, I built AI-driven Smart Home, Smart Building, and Smart City solutions on compact embedded Linux computers paired with trained models — regulating environmental controls, heating/cooling, shades, doors, traffic lights, parking, and more. Positioned as energy-efficient smart systems rather than “AI,” the reception was overwhelmingly positive.

Starting from two people, we grew to 40 within a year, with 300% YoY revenue growth peaking near €1M. Then an unprecedented financial crisis stifled investment; unable to retain and attract customers, I made the difficult decision to close and sell the product. The venture gave me the confidence and expertise to lead any company through scaling, adaptation, and market-aligned product design.

Many Croatian schools lacked IT equipment — projectors, e-boards, PCs, tablets — and there was little learning content to present. The government issued public tenders for full solutions: equipment, software, content, installation, and training.

I crafted a solution using the latest technology, like e-boards and matching content. The hardest part was logistics: 10,000+ PCs, 400 e-boards, plus projectors, speakers, mice, and routers — delivered to 300 schools nationwide, including islands, in under 90 days from contract signing. Risks ranged from 10,000 PCs arriving by ship from a China factory to packaging per-school kits and delivering them a day or two ahead of the field teams (21 vendors, 120+ people).

As Project Manager I owned it all — vendor management, logistics, budget and payments, team and customer management. The government re-ran the tender annually, and I managed these projects for three years, each a success in logistics, deadlines, and customer satisfaction.

Tisak, a dominant Croatian retailer, ran 850+ newspaper kiosks nationwide. Their logistics and kiosk operations were entirely paper-based, causing significant losses from the lack of real-time goods tracking. As Senior Project Manager at King-ICT, I led the digital transformation: conceiving the solution, presenting to the client, contracting vendors, procuring materials, and scheduling the rollout across all 850 kiosks.

Each kiosk got IBM touchscreen POS systems and a mobile router (EDGE, later 3G) for continuous communication with HQ. Warehouses were integrated into an ERP for vast process automation. A machine-learning app optimized each kiosk’s product offering and just-in-time replenishment, surfacing top sellers so operators could transact in three clicks. Goods distribution ran automatically off-hours, and real-time data let HQ monitor operations and decide efficiently.

Over 150 people delivered the solution; ROI was realized in under six months from operational savings. Tisak returned to profitability and was acquired by Croatia’s largest retail conglomerate that year. Today it’s a digital hub — tickets, MoneyGram, bus tickets, photo printing, parcels, card cash withdrawals — all rooted in the solution I introduced.

In 2006 Zagreb — nearly a million residents — relied on a dated public-transport fare system across trams, trains, and buses, using age-old ticketing and validating machines, paper tickets, personalized subscription booklets sold at only two locations on specific days, and a 70-year-old tariff scheme.

At King-ICT I was asked to conceptualize a cutting-edge solution and compete in the public tender. I toured numerous European cities to study their transport systems and, inspired by London’s Oyster Card, devised a three-part solution: a central data center overseeing 4,000+ validating devices across 800 vehicles; in-vehicle Linux color-touchscreen devices on an IP network; and dedicated software on devices and servers for seamless data exchange, replacing paper tickets with contactless Mifare chip cards.

The proposal won. As Project Manager I led a team that grew past 200, running implementation around the clock in three shifts, seven days a week, for over a year, while other teams built the data center and software. We significantly streamlined the tariff system, and the travel data let the city optimize resources dynamically — generating savings that offset the project. Seventeen years on, the system still runs flawlessly with near-zero downtime.

Učilica was a beloved game that made learning from school textbooks fun. Through 2005 it was all 2D, built by a sizeable team of ~15 artists and coders, a couple of leads, and sales staff.

I joined as Senior Developer — no 3D game-making background, but strong at math and building real projects. The team was stuck moving to 3D, especially getting characters to move without glitching; with the math, I fixed those tricky parts in a couple of weeks. Seeing broader disorder, I pitched Agile — and was made project lead, then CTO.

I got the team aligned, working Agile, pairing on tasks, and skilling up on 3D together. Eight months later we’d turned the 2D game into an epic 3D world, right on time for the new school year. It was a hit — sales jumped 300%.

I began as a freelance developer at a young age, first optimizing warehouse operations for my father’s business and then moving into full-scale business process automation. These early wins built my confidence and let me replicate similar solutions for neighboring businesses.

Curiosity pulled me deep into emerging tech and languages — Cobol, Visual Basic, C, C++, Fortran, Visual FoxPro, and Clipper — and I delivered 30+ bespoke automation projects for small enterprises. An early digital adopter in Croatia by 1992, I moved naturally into web development, crafting 50+ websites for hotels, restaurants, and corporations.

In 1997 I wrote a high-school thesis on the then-experimental “Web 2.0” technologies, culminating in a working website with backend integration and dynamic content — an approach that later underpinned social networks, Google, Facebook, and countless SaaS products.