Designing and building the infrastructure, backend systems, and public API that power Browser Use Cloud — a platform running millions of autonomous web agents.

➡️ Architected the sandbox isolation system from scratch with a control plane approach (see blog post)

➡️ Built and unified our AWS infrastructure from scratch, scaling from ~10 to thousands of concurrent agents

➡️ Designed event-driven pipelines for webhooks, scheduled agent runs, and async task processing

➡️ Built the public REST API with multi-model LLM support, file handling, and usage-based billing

➡️ Set up full observability using Datadog to debug the hardest production issues

Helping clients turn ideas into solutions using technologies like:

➡️ Golang (Gin)

➡️ Python (FastAPI)

➡️ Postgres

➡️ Kafka

➡️ AWS

As part of the Operational Finance team I helped build a near-real-time internal accounting software for P&L visibility for freight shipments.

Used Kotlin, Spring Boot, MongoDB and Apache Kafka to:

➡️ Import, transform and store data tied to shipments, routes, invoices, bills and clients in a way that is meaningful for accounting and financial visibility

➡️ Build new features and simplify logic on top of existing micro-services

➡️ Improve our ability to recognise Flexports revenue more accurately.

I also investigated why data is incomplete/incorrect and identified new opportunities for improving data quality in our accounting system.

Moreover, I managed cross-team collaborations since we listened to AWS Kinesis streams from a lot of different teams in the United States.

Lastly, I helped improve the teams onboarding process.

Used JavaScript, Node.js and MongoDB to:

➡️ Design and build new microservices and integrate 3rd party software,

➡️ Create serverless components and core functionalities supporting the event driven architecture using AWS SSM, SNS, SQS, Lambda, S3 and DynamoDB,

➡️ Monitor, fix and upgrade microservices in production

While building and monitoring production software I also interacted with AWS IAM, CloudWatch, CodePipeline, X-Ray and CodeArtifact. I also helped junior engineers with onboarding and mentored them.

Additionally, used Python and AWS SageMaker to:

➡️ Build various pre-processing scripts (local data pipelines) for raw user data and preparing it for our machine learning models, visualisations and user dashboard.

➡️ Build multiple machine learning models (age clocks) from carefully selected biomarkers to present biological age of our users based on their unique biology.

➡️ Retrieve and clean, discover and visualise biological data from external databases to help understand and build longevity scores.

*My first challenge at NU was to use ReactJS and MaterialUI and build our first PoC dashboard for user data presentation.

Worked in the Laboratory for Algorithms and Data Structures at the University of Ljubljana, Faculty of Computer and Information Science, to build an algorithm and user interface that optimizes and automizes the process of fitting different sets of screws in batches. (solving an optimization problem).

➡️ Created an algorithm that sped up the company's process and 𝘀𝗮𝘃𝗲𝗱 𝗮𝘁 𝗹𝗲𝗮𝘀𝘁 𝟭𝟱𝗺𝗶𝗻 𝗽𝗲𝗿 𝗱𝗮𝘆 and got an estimated 𝗶𝗻𝗰𝗿𝗲𝗮𝘀𝗲 𝗼𝗳 𝟭𝟬% 𝗶𝗻 𝘁𝗵𝗲𝗶𝗿 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻. It is written in Java using Maven and Excel documents as input and output files.

➡️ Created a prototype model that predicts the number of trays an order of screws will fill. Used python and the libraries: Pandas, Numpy, Seaborn, Sklearn to create a Neural Network and Linear Regression model.

➡️ Created an algorithm that plans which orders need to be filled in trays first in order to get the best possible outcome of fitting them into batches later.

Additionally, worked with Additherm Group to 𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗲 𝘁𝗵𝗲𝗶𝗿 𝗽𝗿𝗲𝘃𝗶𝗼𝘂𝘀 𝗿𝗲𝗽𝗼𝗿𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺 that involved sending weekly and monthly emails with differently templated excel files.

Since this was highly inefficient we used simple tools and created a new 𝗰𝗲𝗻𝘁𝗿𝗮𝗹𝗶𝘇𝗲𝗱 𝘂𝗻𝗶𝗳𝗼𝗿𝗺 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗿𝗲𝗽𝗼𝗿𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺. Utilizing Google Forms and Google Scripts we created:

➡️ Standardized forms (questionnaires) for collecting data from all companies

➡️ scripts to process and save data from the forms

➡️ Google Sheets that automatically update and can be used for further data analysis

➡️ scripts for automatic email sending to notifying managers to report their KPIs