Tillgänglig ASAP
(Uppdaterat 2024-12-13)Expert Python, C#, Go and Elm Developer
Stockholm, Sweden
Modersmål Hungarian, English
- 24 years of experience in Python and SQL
- 10 years of experience in C#, C/C++ and JavaScript
- 2 years of experience in Go, Elm and Java
Kompetenser (63)
English
OBJECT ORIENTED PROGRAMMING
C
PROBLEM SOLVING
Python
Developer
SOFTWARE ENGINEER
Backend Development
SOFTWARE DEVELOPMENT
Elm
MySQL
Web development
Linux
C++
Assembly
JavaScript
TEAM LEADERSHIP
NUMPY
SQL
INTEGRATION
DATA COLLECTION
TELEMETRY
Linux Administration
Windows
Linux kernel
DESKTOP APPLICATIONS
C#
PostgreSQL
REST API
OPTIMIZATION
Linode
Bash
STRUCTURED SOFTWARE
Go
Frontend Development
PROTOTYPE
MS SQL
MSSQL
CI/CD
PIPELINE
CSS
TENSORFLOW
DevOps
FORECASTING
AI
PySide
Discrete Optimization
Linux Kernel Driver
Azure
Project Management
Functional Programming
ERP
Amazon EC2
WPF
API design
PANDAS
KERAS
Machine Learning
Docker
Oracle
Java
SOLARIS
MAC
Sammanfattning
Senior software developer with 20+ years of industry experience. During my career I gained experience in a diverse set of technologies while developing products in multiple programming languages for Windows, Mac, Linux and the Web.
Professionell bakgrund
2020-03 - 2024-09
Responsible for stabilizing and modernizing Automated Fair Collection systems multiple public transport companies inside Sweden and other European countries depend on for most of their income.
Public transportation is considered as critical infrastructure and has stringent documentation, testing and deployment requirements, especially when it comes to essential infrastructure. Their dependency on legacy hardware and software makes the development, modernization and security patching of such systems non-trivial. Keeping up the high availability requires careful consideration of the risks involved while making engineering or operational decisions.
Viktor had been advising the company on how to transition critical software infrastructure away from legacy technology and to take advantage of modern cloud infrastructure wherever it is allowed and beneficial in practice. He had been actively taking part in developing, documenting and deploying software updates. His responsibility included developing new software and patches for both the embedded frontend devices and the backend server environments they are connected to. All these with strict documentation requirements, change management and security controls.
He had been very successful at the quick investigation and resolution of production issues, including complete reproduction in testing, detailed documentation, mitigation by hotfix and subsequent planned releases. He prefers understanding any software issues in detail before making changes and strives for a comprehensive solution preventing future regressions of the same problem category.
Viktor had been working with large backend databases and their related data processing pipeline. He developed a prototype to generate SQL from free text prompt using a Large Language Model (LLM) running on vLLM for use by BI departments of transport providers and for telemetry dashboard development.
2023-03 - 2024-02
In early 2023 OpenAI released GPT-4 and opened up their quickly expanding catalog of ChatGPT plugins for the public, an entirely new market segment. It was a rare business opportunity, which I could not skip.
Viktor extensively studied the brand-new world of Large Language Models (LLMs) in his spare time, not only learning how to run them efficiently on own infrastructure, but also started to use them for his consulting work and startup activity in different ways.
He developed and released the AskYourCode ChatGPT plugin to the public. It was a RAG solution providing GPT-4 with the requested details about a software project based on a carefully generated code map stored in an SQL database and documentation stored in a vector database. The second production version peak at about 700 daily users.
The project was closed in early 2024, because developers had access to better alternatives (IDE integrations) by then, which significantly reduced the demand for such a plugin. It has been good learning, and the code may still be reused for LLM based projects later.
Viktor has also developed the prototype of an LLM based coding assistant, which can solve small issues detected by static code analysis tools. It has been working reasonably well with local language models, but it did not provide enough value to be marketable, therefore the project had been halted.
2019-07 - 2020-02
Viktor used his spare time between contracts to learn new programming languages, frameworks and tools.
He developed the OptiTax online calculator or small AB companies in Sweden. It allowed them to plan how much salary and dividend to take out to avoid overpaying taxes. It was a good project to learn about modern CI/CD pipelines, container-based deployments, Go as a backend language and Elm as a functional reactive frontend language.
2017-11 - 2019-06
Developed network emulation infrastructure for Microsoft Teams and Skype test labs, leading a small team of developers. This solution was based on former work done by Viktor while he was a full-time employee at Skype Sweden AB.
The custom server infrastructure developed allows for the test cases to dynamically configure the network of the devices the automated calls are placed on. The network conditions may vary for each test case, including the emulation of various bandwidth limits, congested Wi-Fi and mobile network, packet loss, duplication and corruption, faulty IPv6 and NAT64 implementations. The solution also allows capturing the network traffic to help in the analysis of failed test cases.
Microsoft has been using this same solution with thousands of mobile and PC test devices in their test labs in Redmond and Tallinn.
2012-09 - 2017-10
Developed an automated test environment for Skype’s bandwidth manager component, including build configuration, Linux based network emulation, data collection, validation of test results and automation lab management.
Built parts of Skype’s telemetry pipeline, collecting quality data about millions of Skype calls each day. Built internal tools and data pipelines for data-based decision making and tracking of KPIs based on call quality data. (C#, Python, SQL)
2008-02 - 2012-07
Managed a small team of Web, C++ and Python developers working on desktop applications and Web services. Revamped the large Python code base of a game client and released it for Windows, Mac and Linux.
Developed a custom incremental build system to allow for low iteration times. Built a cluster of Linux servers running production and development services.
Developed an online chess board scalable to 4000 viewers/process with 2010 tech. It has been battle tested with 16000 viewers online at the same time during a big chess event. It was cost effective to host, because a small cluster of Linode VMs scaled up only as the demand increased. It used haproxy as a load balancer.
Gained experience in small team management and the recruitment process by hiring talented developers using a straightforward process.
2007-08 - 2012-07
Ghana VAT and MOFEP projects, integrated external systems and thousands of mobile POS terminals. Extended mxODBC Connect to support more SQL servers.
2006-01 - 2009-01
2000-01 - 2006-01
Administrative system to collect data on Hungarian industrial parks, a governmental project. Unified adapter for HL7 compatible medical devices. SMS gateway and various data conversion tools for the K&H Bank. Data recovery tool for a POS terminals.
1998-01 - 2000-01
iNGENICO POS applications, interfacing proprietary hardware. Took part in the development of the official student card system of Hungary. Implemented copy protection scheme for server software. Implemented systematic printing of club cards based on a member database. Maintained the food card system used by the workers in the Sony factory at Gödöllő (Sony Hungaria).
1994-01 - 1998-12
Medical CAD system for flatfoot diagnostics and inset design. This was the first software I sold for a significant amount.
The Windows version is still in use today: https://jantobetet.hu
Uses x86 assembly code for image processing and compression. The initial version was written in Borland C++ for MS DOS and used a video digitizer card for image input. Later the software was ported to newer versions of Windows, with the 3rd version running on Windows 11.
Akademisk bakgrund
1995-08 - 2002-06
Certifikat
2012-10
2015-08
2015-05