Vaidotas Kanopa

I developed an AI-based customer and prospect email request classifier using Salesforce Einstein and co-developed an autoresponder in Salesforce to generate responses.

The problem:
Due to the nature of the business, there were a lot of unanswered prospect email requests posing a lost opportunity, and the costs of treating customer requests were generating huge costs.

Project details:
• Aggregated and transformed Salesforce requests into ML-ready datasets
• Provided the POC version using Python and TensorFlow/Keras
• Trained and tuned models on Salesforce Einstein
• Helped develop an adjustable Salesforce interface for the autoresponder
• Implemented reports to track the performance of the system
• Helped steer the project to achieve the most business value

Result:
The system handled 1,000 to 5,000 prospect and customer requests daily, helping the business improve profitability and increase customer satisfaction.

I helped design the underlying architecture and implemented ETL and Microsoft SSAS migration from local data centers to AWS Cloud.

The problem:
Due to increasing data needs, local data centers became too costly and hard to scale.

Project details:
• Helped design and choose the correct infrastructure for BI needs (EC2 and RDS servers)
• Helped design and implement migration steps ensuring continuous, uninterrupted work
• Designed database schemas and implemented BI data migration to Redshift, RDS, and EC2 Microsoft SQL servers
• Created new adapted Talend ETL flows

Result:
All BI infrastructure was migrated to AWS Cloud, improving the cost and speed with virtually no downtime during migration.

I helped design and implement data democratization in the company, aggregating data from data silos into the central data warehouse and developing Power BI dashboards for analytics.

The problem:
Different departments in the company were using their data and analytics solutions, causing data inconsistencies, lack of data sharing, and inefficient analytics.

Project details:
• Aggregated dispersed data sources into a central data warehouse (AWS Redshift and Spectrum)
• Helped develop a structured Power BI dashboarding system
• Converted into Power BI and optimized existing business analytics solutions
• Co-developed new Power BI reports to answer business needs

Result:
Most of the company's data has been stored in the central data warehouse as the source of truth; data analytics needs have been served by systemized Power BI reports.

I developed an app to find defective wine bottles through visual inspection using image transformation and analysis techniques.

The problem:
In many wine manufacturing plants, defect detection is done manually, which has become costly and introduces human errors.

Project details:
• Based the app on Python and the OpenCV library
• Integrated with industrial-grade cameras
• Achieved real-time speed at 20 evaluations per second
• Developed to be adjustable to new products while evaluating defects against reference images
• Utilized a Kivy-based GUI

Result:
The solution had a human-level performance on most wine bottle types and could process up to 20 evaluations per second.

A classified ads website for selling and buying local businesses.

The problem:
Available options to sell or buy local businesses were limited and lacked functionality.

Project details:
• Created a fully serverless implementation using AWS services (Lambda, Amazon API Gateway, Amazon Simple Storage Service (S3), DynamoDB) and Cloudflare as a content delivery network (CDN)
• Enabled automatic scaling
• Developed a reactive single-page application (SPA) front-end using Vue.js
• Constructed user authentication using AWS Cognito

Result:
A fully functioning serverless scalable classified ads website for selling and buying local businesses.

I developed a flexible WooCommerce-based eCommerce store for a meat products manufacturer that started a food delivery service.

The problem:
During COVID-19, in the local city of the manufacturer, people were struggling to order food and basic products safely due to restrictions on movement.

Project details:
• Created a fully functioning WooCommerce online store
• Developed flexible and adjustable delivery time constraints
• Built custom pricing settings on internal product parameters
• Handled sales reports generation

Result:
The company had a fully functioning eCommerce store under a week from the idea's inception.

I developed an app that continuously searched for potentially profitable products to launch on the Amazon marketplace by scanning trending products and determining the competition.

The problem:
It was manually challenging to find which products were popular and had little competition on the Amazon marketplace.

Project details:
• Created a Python-based app using the Selenium framework for web scraping
• Used the results from Google Ads and other third-party providers to find trends
• Used the results from Amazon to find competition for products

Result:
The app results led to investing in products that became high sellers across all EU Amazon marketplaces.