This page contains list of the projects I have been involved with.
You might be interested only in:
See also my CV and how to hire me as a freelancer.
A web app to optimize 2D package layout for trucks. It provides a simple UI to enter package details and then construct the load by drag-and-drop. Load construction is made easier by using colorful labels and backgrounds. The application is used by Lasita Maja AS.
The package data input area and the work area:
A set of proprietary tools for controlling specific desktop apps with hardware connected through Raspberry Pi. The project contains:
This application started from a weekend project to automatically detect bot visitors and create corresponding filters for Google Analytics (GA). However, at some point of time I started to use it as the main visitor tracker on this blog and some other sites. It allows to browse individual visitor sessions, something that is not possible on GA. Due to not being a widespread solution, it also attracts a lot less referral spam. The code is Open Source and available on GitHub.
List of session and a single session view:
Simplest single-line chart library. Draws the chart using the HTML5 canvas. I built it as I needed something fully responive that I can easily drop into any project that needs simple line charts.
Rendering output:
DOM-EEE is a small Open-Source declarative portable library for extracting data from web pages. I wrote it to simplify development of web scrapers. The library can be used in many JavaScript environments: browsers (userscripts), PhantomJS, Node.js (through Cheerio or jsdom).
A frontend web application to browse and monitor logs stored inside a MySQL database. The application started as a proprietary solution to monitor various projects but later I released it as Open Source. The application replaced my previous solution using a SaaS service that became a bit too pricey for my logging volume. The system also monitors log events from uptime, domain expiration, and SSL certificate expiration check scripts.
An automated system to place sports bets using data gathered from specific sites. The system scraps data from sites and places bets using the API provided by Pinnacle, an online bookmarker. The data on the system is stored in an optimized MySQL instance, running queries on 100k+ of rows in 1-2ms. Part of the system was implemented in SWI-Prolog.
Internal electronic voting system for the Estonian Free Party (Vabaerakond). Includes a voting interface with various types of questions for voters and an administration section to manage question, to manage party members, and to see report of votes on the current questions. The system is intergrated with the cryptographic Estonian ID-card and Mobile-ID APIs to support secure voting.
Scale Laser is an online CAD application and e-commerce site for model railroad builders. My role in the project was to build the overal frontend part of the site, the e-commerce system including the cart and payment integration with Stripe and the backend manager/scheduler for production. The first release is still in development.
22rolling is a vehicle sales portal. It supports 10 different vehicle types and is localized into 8 countries and 9 languages.
Personal cycling and running blog. The blog is built onto the Blog-Core framework. The site design is also created by me.
Photo gallery generator. I did not find gallery generator that satisfied all my needs and I had to build my own. This one supports nested albums, incremental building, metadata in JSON, custom templates (EJS) and generates gallery output into a separate directory tree from the source directories. It is built as a Node.js command-line application and uses ImageMagick to generate thumbnails.
Journaled in-memory JSON database for PhantomJS. The data file format is compatible with Node.js package dirty. The library simplifies development of web scraping code by not requiring external database in projects where it is not needed.
Helper library to make it easier to build KnockoutJS-base Single-Page-Applications. Offers more flexibility than built-in KO components and prevents unnecessary reflows.
A traffic bot to generate real-looking traffic for a specific site using the headless scriptable browser PhantomJS. The bot starts from a random referral page and visits random links on the site. The amount of generated traffic depends on the number of the launched parallel processes of the bot and the configured delay between selecting the next navigation link.
Personal homepage, blog and portfolio page.
Company homepage for Infdot OÜ. The site has been a test ground for various web technologies and has been rewritten many times.
Blogging/CMS framework. Fills the gap in Prolog web frameworks. Comes with a set of simple themes.
An HTML templating library for SWI-Prolog. Installable as a pack.
A data structure validation library for SWI-Prolog. Inspired by JSON-Schema but also includes support for Prolog terms. Installable as a pack.
HTTP router for SWI-Prolog. Uses path expressions (like post/show/Slug) containing logical variables to denote URL paths. Simplifies web application development in Prolog. Installable as a pack.
Sorts list of dicts by a key in dict. Installable as a pack. Recent SWI-Prolog releases contain the sort/4 predicate which obsoletes the library.
An in-memory document-based transactional database for SWI-Prolog. Used as the database in the Blog-Core project. Installable as a pack.
A Markdown parser for SWI-Prolog. Installable as a pack.
A Bash shell script to clone/pull user's GitHub repositories.
An implementation of Promises/A+ by translating the spec directly into code. Passes the test suit but is quite slow. Done for fun.
AVR MCU-based cooling system for PC. The project includes a firmware for AVR and a setup/configuration utility for PC. Built for evaluating an AVR MCU and the avr-gcc compiler. Also done for fun. Deployed to provide silent cooling for my main computer.
A web scrapper for extracting timetables from homepages of UK cinemas. The project included custom web scrappers for about 20 different sites using the scripted headless browser PhantomJS. It also included a Node.js-based web app to store the extracted data in a MySQL database.
A web scrapper to extract sports betting data from the BetExplorer site. The project included a PhantomJS-based scrapper to extract the data and scripts to calculate the required statistics from the extracted data.
An online poker news and affiliate site.
A network of online gambling affiliate sites to serve as landing pages from other marketing channels. Built onto WordPress with custom design and theme.
Ledlamp.ee is an online web store for LED lamps. I inherited the project and did not build the initial version. I helped to relocate the project to the new hosting provider and fixed issues with frontend HTML and PDF invoices.
Logi.ee is a web-based diary for physical activity. My role in the project was to be a technical mentor. I contributed the backend code for integration with the Fortumo and Sendsmaily services.
Continuous Integration script to run project build/test and write result into a log. Supports Git only. Works in cases when Jenkins is overkill. The build log output is supposed to be monitored by a realtime logging service.
Attempt to document the offline database synchronization system developed for the FitSmarty project. The code sets up a MySQL database and two SQLite databases. Then runs random operations in them, synchronizes, and checks consistency. The code is very verbose but works.
Client-side hashchange regex router for singe-page applications. Works on Chrome 5, Firefox 3.6 and IE 10+. Initial codebase was extracted from one of my frontend projects. Since then it has been improved with methods required for other projects. The library includes support for asking confirmation to leave the current route. This feature was not supported by other front-end router libraries.
Small library for making REST queries. It takes away the most verbosity of using XMLHttpRequest and has hooks for request start/end. It is now rendered obsolete by the browsers new built-in fetch API and the Promises.
Keha Arengu Monitooring (KAM) is a computer software to track person's health and fitness. It was initially based on the popular Excel worksheet distributed at tervislik.net. The objective of the project was to implement the application which would be more usable, more maintainable, and with more features than the Excel worksheet. FitSmarty was the English localization of the application.
The application supports all 3 desktop platforms: Windows, Mac OS X and Linux.
The web/mobile port of the KAM/FitSmarty project.
Included custom offline data replication between MySQL master and SQLite slaves.
Custom CRM to handle sales of the KAM software. It includes the frontend store view and the backend with user management, sales, license key management and invoicing.
Includes integration with Estonian banks to handle payments.
Helper application to track users of KAM/FitSmarty software. It includes both the web application for collecting statistics and displaying reports and the client code to send application usage events from the desktop software. The application took 3 days to put together.
Personal web-based replacement for Google Reader and other feed readers. It tracks seen/read/important state of articles and is built for high performance. Scales well beyond 500 feeds that the most other feed readers struggle with.
Node.js binding for the RapidXML parser for very fast newsfeed parsing.
The Estimator is an application for estimating the time that a project takes. It divides the project into components and components into tasks. The tasks are estimated to take min to max hours. The app then calculates total min to max hours for the whole project.
The application was built to evaluate the KnockoutJS and Bootstrap frameworks.
Uses the localStorage API to store data.
Proof-of-concept source-to-source Prolog to JS compiler. Includes a translator written in Prolog and a small runtime (200 loc) written in JavaScript. It can run simpler programs, like the queens benchmark.
A RESTful API for entering time logs remotely. The API uses HTTP GET/POST with JSON-based messages and error handling. My scope of the work was writing the initial API specification.
A Prolog library to parse YAML documents. Supports only a good subset of YAML.
A double-entry financial journal. Keeps records in YAML files.
Sitecheck is a Bash script to monitor a number of web sites. It periodically queries the list of sites and logs the OK/FAIL result. The log itself can be monitored using a real-time logging service.
Specification and code for the thread analysis (main only, unique, many) of the C language for the Goblint static analyzer. The thread analysis was supposed to reduce the number of false positives from the thread locking/synchronization analyzer.
A toy project to study the problem of dataflow analyses during the time I prepared my master thesis. The code includes many common analyses used by compilers.
LLJ is a Prolog-like logic language for running on JVM. It allows close integration with Java libraries. The source code is extensively commented and covered with a high number of test cases. At the time of release it was faster and more memory efficient than many other Prolog implementations on JVM.
A Redmine plugin to automate release file management. The plugin allows the build artifacts to be uploaded into Redmine File's section using curl or a similar HTTP utility.
A project to provide fast and accurate search results from a collection of existing documents. The project also included PHP scripts to dump documents into the search database. The search application was tuned to retrieve the search results from over 100000 documents in less than 0.1 seconds.
The WordNet database was used for implementing synonyms support.
An online pool (8-ball) game. The game features game rooms, chat and pool game between 2 players and observers.
Ignition starter for vehicles with internal combustion engines. The project included custom circuit design and firmware.
Part of a coursework to implement a compiler for a lazy functional language called Puf. The project implemented compiler in the Java language. I worked together with Merli Uus. My contribution was mostly the parser and the optimizations.
My old blog engine. Written to evaluate some new (at the time) Java technology. It never got any posts.
The Seals Database for the National History Archive. My responsibility was to gather requirements from the archive specialists, document them, produce the application specification, and later test the actual application against the specification. The set of acceptance tests was implemented using the Selenium test utility.
A web-based front-end for mplayer. It extracts metainfo (title, artist and album) from music files (mp3 and flac) and makes it easy to create playlists based on this info. The metainfo is extracted using the entagged library.
Self-service portal for Elion customers. Elion is the largest telecom company in Estonia. I worked in a team of 6 people as a Java programmer. My responsibility was to implement features based on the HTML prototype and specification documents. Large part of the work involved writing PL/SQL unit tests for the legacy Oracle database.
The repo with ancient code going back to 2001: https://github.com/rla/old-code
