Vadim Ovcearenco

Developed back-end services for an AI-powered platform that generated custom aquarium layouts using Stable Diffusion. Implemented secure authentication with Auth0, built REST APIs using Django, and managed PostgreSQL databases deployed on AWS. Integrated RunPod GPU infrastructure with ComfyUI by implementing a workflow execution pipeline based on JSON graph definitions, enabling automated image generation from user requests. Collaborated closely with the front-end team to deliver an end-to-end experience from prompt submission to AI-generated aquarium designs while ensuring scalability, reliability, and maintainability of the back-end services.

Developed back-end services for an insurance document automation platform that generated legal documents from structured data. Implemented a template engine based on Microsoft Excel templates, enabling business users to define document layouts without code changes. Built Python services to transform application data into finalized insurance documents, improving automation, consistency, and maintainability of the document generation process.

Developed and maintained a commercial real estate financial modeling platform built with Django, MySQL, and AWS. Served as the primary back-end engineer, implementing new features, resolving production issues, and working directly with the client to deliver business requirements. Extended the platform's Excel-based financial modeling engine using PyWin32 and COM automation, enabling dynamic generation and customization of complex financial models based on user input. Improved the maintainability and reliability of the application while supporting its transition from the previous developer and ensuring smooth day-to-day operation.

Massive multiplayer mobile project that involved many different disciplines like ultra low latency live streaming video, Elixir web sockets and a Django back-end for CRUD tasks and web views.

Role:
System Admin, DB Admin, Web Admin, Back-end Python and Elixir Developer

Technologies:
AWS, NGINX, MySQL, Python, Django, JavaScript, Elixir, Phoenix

Challenges:
We used Elixir/Phoenix to create web socket channels with clients. It gave us best performance and almost no latency.

Prymesyft is software house specialized in web applications. They needed the app to control contracts, payments, project progress and tasks management for the client.

Role:
Full-stack Developer

Technologies:
AWS, EC2, RDS, Python, Flask, SqlAlchemy, AngularJS

Challenges:
Find the best AngularJS architecture to provide easy development and updates.

The application to archive email/web data of a client, where the client can create account on the web app and set up what site and what mails he/she wants to archive, and as for the tech stack the folowwing is used, Python, Django, Docker, Amazon SQS, Amazon SNS, Vue.js, Postgres, Elasticsearch.

Role:
Full-stack Developer. I was responsible for email archiving.
All incoming emails should be parsed, inserted in DB and search engine (Elasticsearch).

Challenges:
Create a robust workflow for email processing. User permission access, Effective archive search and management.

Description:
Collecting, processing and analyzing skin photos using AWS. The company previously had the project running on GCP.

Role:
As most of the code was written for GCP (by me), I had to migrate to AWS using AWS tools.

Technologies:
EC2, Elastic Beanstalk, Elastic Container Service, Lambda, Auto Scaling, Docker, S3, RDS, SQS, SNS, CloudWatch. CodeCommit, Flask, Ubuntu, JavaScript

Challenges:
Find the best tools in AWS to optimize performance vs. cost. Improve the error handling flow.

Description:
Collecting, processing and analyzing skin photos using Google Cloud Platform (GCP)

Role:
I had to design software and hardware architecture to create an API REST interface for the mobile client, to create a minimum-functionality web client using JavaScript libs.
The system should be able to process 6000 photos per hour.

Technologies:
BigQuery, Storage, Datastore, Kubernetes, Google Logging, Compute Engine, Queues, Flask, Ubuntu, JavaScript

Challenges:
To provide required performance, we used Google Container Auto-scaling. All errors are saved in a special queue for processing manually later. The analyzer was run every day and computed all statistical data.