Senior software engineer, Brno

Software that plans routes, reads emails and saves people manual work.

Since 2019 I've been building Wereldo's logistics platform from the ground up, owning features end to end on a small team: Vue on the front end, Go and Node.js services, AWS underneath — and AI where it actually pays off.

Open to senior full-stack and backend roles

stops
22
vehicles
4
distance
1,912 -73%
A toy version of what my production engine solves: which vehicle serves which stops, and in what order. This one runs a simple heuristic (Clarke–Wright savings plus 2-opt) right in your browser. Click the map to add a stop.

Selected work

Most of what I've built lives inside a product I can't open-source. Here's what it does and how I approached it.

  1. Wereldo · 2020 – now · lead engineer

    Route optimization engine

    Dispatchers used to plan routes by hand, assigning shipments to trucks and ordering stops from experience. I designed and built the engine that does it for them. It treats the job as a vehicle routing problem with pickups and deliveries, respects time windows and vehicle capacity, and proposes a complete plan right inside the platform. The solver grew out of my master's thesis, and I then hardened it for day-to-day production use.

    Plans are generated automatically and no longer depend on who's on shift.

    Go · AWS · combinatorial optimization

  2. Wereldo · 2023 – now · lead engineer

    LLM pipeline: from quote email to order

    Quote requests arrive as free-form emails, and someone had to retype each one into the system. I built a pipeline that extracts a structured order from every email and hands it straight to the sales flow. The hard part wasn't the demo, it was making it reliable: agentic evals that catch regressions before release, monitoring of output quality and latency, and prompt and model tuning (open-weight models included) to keep inference costs down.

    No more manual retyping, and quotes go out faster.

    TypeScript · Node.js · LLMs · evals · OpenTelemetry

  3. Wereldo · 2021

    Route plan forecasting

    A route plan is just a list of stops until reality kicks in. I added stochastic risk evaluation on top of the plans, so dispatchers can see how likely a plan is to hold up before they commit to it, not halfway through the day.

    Go · PostgreSQL · AWS

  4. Wereldo · 2022

    Invoicing automation

    Getting from finished work to a sent invoice took several manual steps, each one slow and each one a chance for a typo. I automated the whole pipeline, so invoices now go out as a normal part of operations.

    Go · Node.js · PostgreSQL · AWS

  5. Brno University of Technology · 2020

    Master's thesis: freight route optimizer

    Vehicle routing with pickups and deliveries is NP-hard, so brute force is off the table. I compared a genetic algorithm with (μ+λ) and (μ,λ) evolution strategies and added guided mutations that work with route structure and route centroids. Plain C++, standard library only. It became the foundation of the production engine at Wereldo.

    C++ · genetic algorithms · evolution strategies

Experience

Feb 2019 – present

Senior Software Engineer · Wereldo

  • Designed and built the company's core logistics SaaS from the ground up, including a vehicle-routing optimization engine that automated route planning previously done by hand.
  • Built an LLM pipeline that turns incoming quote-request emails into structured orders, removing manual data entry from the sales flow and speeding up quote turnaround.
  • Ran it in production with a focus on reliability and quality: agentic evals to catch regressions, quality and latency observability, and prompt and model tuning (including open-weight models) to cut inference costs.
  • Developed route-plan forecasting with stochastic risk evaluation and automated the invoicing pipeline.
  • Own features end to end on a small team: Vue.js front end, Node.js and Go microservices exposing GraphQL and REST APIs, AWS infrastructure.

Jun 2016 – Sep 2017

Web Developer · Simplo

  • Back-end-focused full-stack developer building web apps and information systems in PHP and Laravel for a small agency.

Education

2017 – 2020

Ing. (MSc), Computer Science

Brno University of Technology, Faculty of Information Technology · Information Systems

2013 – 2017

Bc. (BSc), Computer Science

Brno University of Technology, Faculty of Information Technology

Tools

Core
TypeScript / JavaScript, Go, Rust, Vue.js, React, Node.js, GraphQL, REST APIs, SQL
Also
Python, PHP / Laravel, microservices, serverless, Linux
Cloud & ops
AWS, Docker, Kubernetes, CI/CD, Redis, Pub/Sub, OpenTelemetry, Grafana / Prometheus, Sentry
Applied AI
agentic evals, LLM observability, LLM cost optimization, open-weight models, model serving, RAG, prompt engineering

About

Michal Gabonay

I'm a senior software engineer based in Brno. For more than seven years I've been designing, building and running SaaS products, from the first line of code all the way to production. I work across the front end, the back end and cloud infrastructure.

I use AI where it makes a real difference for users and the business, and I care about quality, observability and whether it pays for itself.

I got into route optimization through my master's thesis, and shortly afterwards it became the core of a real product.

Languages: Czech and Slovak (native), English (fluent), Polish (advanced).

Contact

Email is the fastest way to reach me. I reply within 48 hours.

Or write here