GenieAI is a portfolio management system for institutional investors investing in Cryptocurrency. GenieAI tracks digital assets spread across centralized exchanges and prime brokers. Besides data collection, GenieAI analyses current holdings to compute financial metrics. Investors could use AI and Mathematical Optimization tools built into the platform to guide future investment decisions.

Genie AI is a cloud-native application deployed in Kubernetes in AWS. The system is composed of a few multi-services glued together using GraphQL federation. Python is the primary language at the backend.

Role

Time Frame

Tasks

• implementing core features on the backend site,
• scaffolding the project,
• setting up AWS infrastructure (i.e., Kubernetes clusters),
• building CI pipelines, establishing the release process,
• managing the dev team, conducting regular 1:1,
• interviewing candidates for SE positions,
• supporting the CTO in feature planning and technical decisions.

Technology

• Python
• FastAPI
• Pydantic
• SQLAlchemy
• Click
• Strawberry GraphQL
• Prefect
• Pytest
• Postgres
• Apollo GraphQL
• Terraform
• Kubernetes
• Docker
• Helm
• ArgoCD
• AWS
• Grafana
• Sentry
• GitHub Actions
• HashiCorp Vault

The customer appointed the software house I worked for to build a content management system for processing legal matters using a modern tech stack. The system allowed legal departments to introduce a formal process for handling matters, distribute the workload across multiple legal professionals, track matter completion in the pipeline, share documents with external partners, etc.

The system was a multi-tenant cloud-native application built in Java and deployed in Kubernetes in Azure.

Role

Time Frame

Tasks

• scoping, designing, implementing, and testing customer-facing features,
• leading a feature team,
• onboarding new customers,
• managing a self-hosted Keycloak deployment.

Technology

• Java 17
• Spring
• Java Persistence API
• Python
• Postgres
• ElasticSearch
• Keycloak
• Kubernetes
• Azure
• Ansible
• Terraform

Technology

• Python
• HTTP

A six-month project aimed to demonstrate the scheduling and routing problems solved by human planners for the Glasgow Home Care Unit could be optimized using existing technology. The secondary objective was to estimate savings available through schedule and routing optimization.

The problem instances solved were the largest reported in the academic literature, and the schedules found during optimization significantly outperformed human planners. I described the solution methodology and results in an article published in the Journal of Operations Research Society.

Role

Time Frame

Tasks

• requirements analysis, design, implementation, and pilot deployment,
• building a suite of sample optimization problems using historical data obtained from the Glasgow Home Care Unit,
• developing a Constraint Programming (CP) model for routing and scheduling of home care visits,
• solving the CP model using Google Operations Research tools.

Technology

• Constraint Programming
• Time Series Analysis
• C++ 11
• STL
• Boost
• CMake
• Google Optimisation Tools
• Open Source Routing Machine
• Python
• Pandas
• matplotlib
• Travis CI

Turbo.net is a system for building, distributing, and installing containerized applications for Windows.

In its primetime, Turbo.net competed with Docker to become the leading container technology for Windows. The company has solved many challenging technical problems for which they secured patents, i.e., to run a container with a click of a button in the web browser and to start a partially downloaded container image while the download is still in progress.

Role

Time Frame

Tasks

• implementing new features and fixing bugs in the command-line client, a desktop application, the server backend, and the language for defining application images,
• building an automated pipeline for creating application images,
• developing open-source plugins for integrating the platform with Jenkins, Vagrant, and web browsers (i.e., Firefox, Chrome, Internet Explorer),
• writing technical articles for the company blog,
• interviewing candidates for SE roles.

Technology

• C#
• NUnit
• Scala
• ScalaCheck
• Lift
• PowerShell
• C++
• JavaScript
• Vagrant
• Jenkins

Role

Time Frame

Tasks

• supervising a team of six remote software engineers responsible for running and fixing E2E tests,
• solving technical problems escalated by the team,
• implementing support for new UI components in the testing framework,
• managing an intern,
• interviewing intern candidates.

Technology

• C#
• Windows Automation API
• PowerShell
• Microsoft SQL Server

Role

Time Frame

Tasks

• designing and implementing an internal Groovy Domain Specific Language (DSL) for SIS configuration. The feature extended the expressiveness of rules operators could define and reduced the framework boilerplate compared to XML-based configurations. My team presented the DSL at the 14th International Conference on Accelerator and Large Experimental Control Systems.
• configuring an SIS deployment for a new particle accelerator (LINAC 4).

Technology

• Java 6
• Spring Core
• RMI
• JUnit
• Groovy

TeleDICOM is a platform for scheduling and performing remote consultations between healthcare professionals. Participants of the video conference could share, review, and work together on medical documents in the DICOM format.

TeleDICOM was implemented in .NET using the Service Oriented Architecture paradigm. System components were deployed as Windows services, and users accessed the platform via a desktop application.

Role

Time Frame

Tasks

• implementing a service for removing information that could identify patients from images, videos, and metadata saved in the DICOM format,
• developing a module for managing remote consultations in the client application.

Technology

• C#
• DICOM
• Windows Presentation Foundation
• Windows Communication Foundation
• NHibernate
• Leadtools Medical Imaging
• Microsoft SQL Server