I'm working on Scala 3, the next version of the Scala programming language.

Compiler Engineer

Secure Code Warrior is a gamified educational platform where people can learn secure web application programming. My role is to compose vulnerable challenges and the solutions to them in Scala Play. I am also integrating the Typelevel stack (Cats, Cats Effect, Doobie) in my solutions.

Software Engineer

For this contract, I am working on a book "Mastering Functional Programming". I am writing about the functional paradigm for imperative programmers. Topics include type classes, Typelevel libraries (cats, doobie, http4s), patterns of advanced functional programming (tagless final, Aux, monad transformers). Topics of asynchronous programming are also included: cats-effect library is discussed and Akka Actors infrastructure is presented.

Author

I prototyped a chatbot for a startup in the travel industry. To do this, I used Stanford NLP Technologies, Python, Scala and IBM Watson cloud solutions. Zoteca Web Application Engineer

Conversation Systems AI Engineer

This project was an online store developed in Scala from scratch. We were tasked to develop a webshop with scheduling capability for an American textile startup. The scheduling functionality was for users to book time slots on store popup events so that they can be measured and custom clothes can be created for them.

I was responsible for developing the store from zero, as a full stack engineer. We used Scala on the backend and ScalaJS on the front end. We employed Typelevel stack and purely functional paradigm (Cats, Doobie, Functional Streams (FS2), Circe etc) throughout the codebase. We used Stripe API for payment integration and Azure cloud services for authentication and data storage.

We were able to maintain the project for half a year, servicing the users through several popup events in process.

Web Application Engineer

The medical insurance industry faces a growing problem over the fraud by medical professionals. They may use various tricks, such as prescription of a redundant drug to a patient, so that to bill the insurance companies more than necessary.

I worked as a data science consultant on fraud detection tasks for a health insurance startup. Based on a dataset sample of the insurance company database, I used Docker, R, PostgreSQL and Scala for data exploratory and fraud detection purposes.

Another direction we worked on was the creation of a chat bot for the insurance companies. The idea was to allow the health practitioners and clients to do everyday tasks, such as submitting the bills of the patients to the insurances. We implemented a telegram bot for these purposes using Scala with Telegrambot4s. We used NLP techniques, such as slot filling, for dialogue management. I worked on a rule-based NLP model to extract the intent of an incoming message. For the server infrastructure, we used Twitter Finagle with Finch as HTTP server. We used Doobie to contact the Postgres database.

As a result, we were able to provide our startup client with a prototype along with intuitive visualisations they were able to pitch to the interested parties.

Data Scientist

As part of this job, I was working on two video courses:

✔️Spark for Data Analysis in Scala - covered usage of Spark on an example of a Kaggle problem using the Spark Notebook in Scala.
✔️Advanced Techniques for Data Analysis with Scala - covered DeepLearning4J/DeepLearning4S for Deep Learning in Scala, Bokeh for plotting and integration of Spark with these tools. To date, more than 100 students took the courses.

PowerPoint presentations are widely used for conference talks, company meetings and even in some karaoke restaurants. It is often the case that slides of a presentation follow a single pattern. However, there are little tools to abstract these patterns, so that the presentation creators define them once and use them to present all the data they have.

In this project, I worked on a template engine for PowerPoint. I wrote analytical tools for Open Office XML using ScalaXB, the XML bindings generator for Scala. The analysis yielded a representation of the template to be processed. The rest of the system defined a DSL the end user could use in their slides to define the presentations. Also, as part of the project, I performed a number of contributions to ScalaXB.

We were able to launch the project in late 2016.

Software Development Engineer

This project aimed at sentiment analysis of Twitter and blog posts, to predict in real time what people like or dislike about a product. Timely knowledge of this information allows production companies to test new products fast, save money on products that perform poorly and increase profits on the promising ones.

✔️I developed machine learning models using Python Tensorflow and Keras for the project. I prototyped Deep Learning models, such as CNN, RNN, LSTM.
✔️I also programmed data mining capabilities using Scala, Twitter API via Twitter4J/Twitter4S and streaming via FS2 (part of the Typelevel umbrella, utilising Cats infrastructure).
✔️Finally, I worked on data exploration and visualisation in Gephi, the graph software. As a result, we were able to deliver the end client a prototype that they used to pitch the idea to investors. The sentiment/semantic accuracy of classification reached 80%.

I continued my work on the implementation of the SubScript extension for Scala. SubScript adds a process algebra syntax to this language. In previous versions of SubScript, we had to use a Scala compiler fork to define the DSL of the process algebra. Usage of a compiler fork created a high entry barrier for the early adopters.

During this experience, I worked on making SubScript more user-friendly by moving the DSL implementation from the compiler form to an SBT plugin.
✔️The functionality from the Parser phase of the Scala compiler was implemented as a Parboiled2-based parser and an SBT plugin. The SBT plugin used the parser to generate synthetic Scala sources from SubScript-containing Scala files before the compilation phase.
✔️The functionality from the Typer phase of the Scala compiler was previously used for type inference. I moved it entirely in macros.

As a result, we were able to run SubScript code from a standard Scala compiler from a standard SBT distribution. This allowed for much easier adoption by newcomers. Previously, they had to fork our compiler and use it without any build tool from the command line. Now, using SubScript is as easy as adding a single line to your SBT build script. See less

Akka actors can be tricky when they involve state. E.g. often you need to change how Akka responds to messages as a result of handling of a message. Normally, this is done following the State Machine model: switch the behavior to another when a message arrives.

During GSoC 2014, I integrated SubScript, the process algebra implementation for Scala, with Akka. As a result, SubScript users were able to describe changes in message handling using intuitive DSL rather than obscure state switches.

Google Summer of Code Intern

I worked for an open-source project that implemented a domain-specific language for Scala to describe asynchronous and parallel software. I have implemented type-inference features for the language via Scala compiler modification and macros. As a result of my work, the users of the DSL were able to leverage Scala's type system to infer the types in their DSL scripts. Self Employed

During my early days in programming, I worked on software development projects in Java, PHP and Unity3D.

✔️I developed a prototype of an AR application for a Singaporean educational startup. The application rendered 3D models over 2D images in textbooks, which allowed for richer exploration of content and gamification.
✔️I worked on smaller projects involving Wordpress plugin development, J2ME mobile applications, Java EE enterprise applications (Servlets, Spring) and Unity3D games for Android.

Independent Software Contractor

I worked with web application backends, Spark for data engineering, purely functional architectures, and Scala.js for the frontend part. Currently, I'm supporting the Scala 3 effort at EPFL.

I've been a core team developer of the Scala 3 programming language for the past 4 years.

I was involved in the development of data analysis applications for the medical and hospitality industry. For those purposes, I was using Python. Currently, I'm working with Langchain to create apps that use LLMs.

I've been working with the JVM platform on the Scala and Java languages for the last 10 years.

I've been using React's GatsbyJS framework for my personal website, as well as SvelteKit to build side project web apps over the past few years.

Personal details

Anatolii K.

Summary

Technical skills

Work Experience