Lasha Kharshiladze, Developer in Tbilisi, Georgia
Lasha is available for hire
Hire Lasha

Lasha Kharshiladze

Verified Expert  in Engineering

Computer Vision Developer

Location
Tbilisi, Georgia
Toptal Member Since
January 23, 2020

Lasha is a software engineer with three years of experience building web apps using Python (Flask, Django) and two years in machine learning and computer vision using Python and C++. He is also a deep learning practitioner and enthusiast. Lasha believes that the key to a successful project is a genuine interest in the product so that the developer doesn't focus on the technical implementation alone but also the final product holistically.

Portfolio

ICardio.ai
DICOM, OpenCV, TensorFlow, PyTorch
MaxinAI
Flask, Docker, Fast.ai, PyTorch, OpenCV, Python, Internet of Things (IoT)...
InvolveAI
Python 3, Flask, React, Scikit-learn, BigQuery, Amazon Web Services (AWS)...

Experience

Availability

Part-time

Preferred Environment

Jupyter Notebook, PyCharm, Windows, Linux

The most amazing...

...thing I've ever developed was a product information extraction service (product name, nutritional table, allergens, etc.) for C-LABS, a food regulatory com.

Work Experience

Computer Vision Lead

2021 - 2022
ICardio.ai
  • Added a View Classification that detects which specific view was used by the doctor when taking a cardiography: A4C, A5C, PLAX, and others.
  • Implemented a Heart Parts Segmentation. It segments Left/Right Ventricle, Left/Right Atrium, Tricuspid Valve, and Mitral Valve in real-time with very high accuracy.
  • Added Calculating Ejectiong Fraction feature. It calculates ejection fractions from DICOM using two methods: 3D convolutional neural networks and LSTM.
  • Added Keypoint Detection and Tracking. This enabled users to detect centroids of the Left/Right Ventricle, Left/Right Atrium, Tricuspid Valve, and Mitral Valve. It also enabled real-time tracking.
Technologies: DICOM, OpenCV, TensorFlow, PyTorch

Computer Vision Specialist

2018 - 2022
MaxinAI
  • Worked on a project for C-LABS SA where we built an expert system that extracts nutritional information from a product box photo taken by the client and checked if it violates food regulations for a particular country.
  • Implemented Jira for task management and Jira Confluence for documentation.
  • Used Scrum methodology for the first half of a year-long project, before moving onto Kanban.
  • Distributed tasks between three junior computer vision developers and three back-end engineers for the C-LAB project where we implemented a website for food product boxes validation/regulatory.
Technologies: Flask, Docker, Fast.ai, PyTorch, OpenCV, Python, Internet of Things (IoT), Machine Learning, Computer Vision, Amazon Web Services (AWS), Amazon EC2, Amazon Elastic Container Service (Amazon ECS), Kubernetes, Big Data

Full-stack Developer | Data Scientist

2020 - 2021
InvolveAI
  • Developed data connectors to export data from different sources: Salesforce, Outlook, and BigQuery.
  • Made predictor of customer churn and growth for list of customers using ML model.
  • Optimized data integration steps and make them asynchronous and multi-thread to export as much data as possible in a limited time.
Technologies: Python 3, Flask, React, Scikit-learn, BigQuery, Amazon Web Services (AWS), Amazon Elastic Container Service (Amazon ECS), Amazon EC2, Pandas, Amazon SageMaker

Full-stack Developer | Computer Vision Engineer

2020 - 2020
Luova Education
  • Developed proctoring app that helps teachers manage online exams and get alerts in case of whisper, multiple face detection, or opening another tab session.
  • Implemented webcam and screen streaming from students to proctor and live chat between them.
  • Created the platform Qwiz and alerts managing board using React/Redux and Materialize library.
Technologies: Python 3, WebRTC, TensorFlow, Video Processing, Streaming, Audio Streaming, React, Redux, WebSockets

Data Scientist | Computer Vision Specialist

2017 - 2018
Pulsar AI
  • Worked as a data scientist and computer vision specialist in Pulsar AI.
  • Developed an optical reader of Georgian national ID with Computer Vision.
  • Built a chatbot engine platform that gives clients the ability to build their own chatbots.
Technologies: Flask, Docker, Pytesseract, OpenCV, Python

Junior Java Developer

2017 - 2017
Fina, LLC
  • Developed a dynamic knowledge management system website for the Ministry of Environmental Protection in Georgia.
  • Built a dashboard page for this application, fixed bugs, and created reports about environmental data.
  • Constructed a dynamic knowledge management system that won the sustainability award in the category of planets.
Technologies: Apache Maven, Ext JS, WildFly, Enterprise Java Beans (EJB), Java

Teaching Assistant

2016 - 2017
Free University of Tbilisi
  • Reviewed and estimated other students' projects in programming methodology and programming abstraction courses. The projects were written in Java and C++.
  • Advised them on how to write better code and make clear documentation.
  • Composed midterms in programming methodology and programming abstraction.
Technologies: C++, Java

Involve.ai

https://involve.ai
Worked on early warning dashboard to predict customer churn and growth. Used AI-based approach that unifies all customer data and can predict who is likely to churn and who is headed to success in advance.
I was a back-end engineer and data scientist; I developed data connectors to export data from different sources: Salesforce, Outlook, and BigQuery. I also made a predictor for "probability to renew" for a list of customers using the ML model.

Qwiz Education Proctoring App

I worked as a full-stack developer and AI (computer vision) and created the platform Qwiz, the total online proctoring solution for K to 12 learning organizations. Secure, easy to use, and a full suite of features designed to give you an advanced, fully customizable testing system that ensures students' and teachers' academic integrity in the online learning environment.

Optical Reader for the Front Side of the Georgian National ID Card

I enabled the automatic detection of an ID card's location in a photo by finding its coordinates, cropping, and reading each field separately. With this product, it is easy to check if a person's identity is genuine by confirming if the fields filled by the client are the same as in the document.
2019 - 2020

Master's Degree in Computer Science

Tbilisi State University - Tbilisi, Georgia

2015 - 2019

Bachelor's Degree in Mathematics and Computer Science

Free University of Tbilisi - Tbilisi, Georgia

Libraries/APIs

Flask-RESTful, Matplotlib, REST APIs, TensorFlow Deep Learning Library (TFLearn), PyTorch, OpenCV, NumPy, Fast.ai, WebRTC, Libav, Reddit API, Google APIs, Scikit-learn, Pandas, FFmpeg, Blender Scripting API, TensorFlow, React, React Redux

Tools

AWS Glue, Amazon Elastic Container Service (Amazon ECS), Pytest, ABBYY, Amazon SageMaker, Amazon Textract, IPython Notebook, Git, PyCharm, WildFly, Apache Maven, BigQuery, Jira, Trello, Blender

Frameworks

Flask, Qt 5, Next.js, Material UI, Django REST Framework, Qt, Ext JS, Redux, Django, Scrapy, Selenium, Swing

Languages

Python 3, Python 2, Python, HTML, HTML5, CSS, TypeScript, JavaScript, Java, C++, C++14, Java 8, SQL

Paradigms

Scrum, API/Services Architecture, RESTful Development, Model View Controller (MVC), Data Science, Unit Testing, DevOps, Agile, Testing, Test-driven Development (TDD)

Platforms

Linux, Salesforce, Amazon EC2, AWS Lambda, Amazon Web Services (AWS), Docker, Windows, Jupyter Notebook, Kubernetes, Azure

Storage

Redshift, NoSQL, Data Pipelines, Google Cloud, Redis, MySQL, PostgreSQL, SQLite

Industry Expertise

Healthcare

Other

Videos, Data Engineering, Audio Streaming, WebSockets, APIs, Back-end, Software Architecture, Data Analysis, Mathematics, Mathematical Modeling, Mathematical Analysis, Computer Vision, Web Scraping, Algorithms, Data Scraping, Internet of Things (IoT), Big Data Architecture, Product Vision, Real-time Vision Systems, Vision Development, IoT Protocols, Big Data, NVIDIA Jetson Nano, Containers, Data, Video Streaming, Object Detection, Convolutional Neural Networks (CNN), OCR, Amazon Machine Learning, Tesseract, Machine Learning, Deep Learning, Image Processing, Video Processing, FastAPI, Software Development Lifecycle (SDLC), Audio, HTTP Live Streaming (HLS), Pytesseract, Streaming, Natural Language Processing (NLP), EJB3, CGI, Chatbots, Scraping, Agile Data Science, Enterprise Java Beans (EJB), DICOM, Generative Pre-trained Transformers (GPT)

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring