Irakli Gugushvili

我们创建了一个渗透测试平台，用户可以在其中运行和监视预定义的攻击. 用户可以手动创建自定义攻击，并对其进行测试、运行和监控. We used Django for the back end, PostgreSQL for the database, AWS for the deployment and separate asynchronous tools, GCP for bot creation and running, and Elasticsearch for logs.

我们开发了这个系统来监控Azure用户的活动，使用内置和自定义策略组合来检查合规性. 在不符合规定的情况下，系统会提醒业主解决问题. We could also do the enforcement on our end if needed. We used Azure Logic Apps to start, Azure Functions to execute the logic, Azure Queue Storage to communicate between functions, and Azure Table to save the data.

This system allowed a root user to monitor the activities of sub-users. 它以一定的时间间隔运行，收集子用户活动，并在需要时强制执行遵从性. We used several lambda functions, SQS for communication, DynamoDB for saving the data, Elasticsearch for logs, Jenkins for CI/CD, and Terraform for infrastructure as code.

这个项目是关于为视频流媒体平台创建一个推荐引擎. I created the entire infrastructure, starting from ETL using Azure Functions, the back-end API using Flask, and finally, vector comparison and prediction using Faiss.

我们创建这个网站是为了将一个具有NLB和TG设置的VPC附加到没有公共访问权限的EC2实例上. Using the server this way was much safer. 我主要关注后端，在AWS Step Functions中实现了整个attach和detaching流水线. We used React with an API Gateway on the front end, Jenkins for CI/CD, and Terraform for infrastructure as code.

这个项目包括使用Scrapy抓取各种公开可用的数据. 我为不同的数据源创建了管道和所有不同的蜘蛛. The main challenge for static websites was parsing different pages; for dynamic websites, it was simulating a JavaScript request that loaded the data. 我们使用Scrapinghub(现在的Zyte)进行部署，并使用带有DynamoDB的AWS S3来保存数据.

In this project, 我使用各种工具创建了一个数据分析器和预测系统，用于不同类型的数据. 然后，我使用Flask-RESTful创建了一个API，并将上面提到的系统包装在其中.

This is a bookmarking system that saves URLs and their locations. 我们通过使用谷歌的地理位置和天气数据来完成这个任务，这些数据是通过IBM文本分析器抓取和分析的. 我们使用部署在AWS Elastic Beanstalk上的Flask作为主要技术.

该系统可以接收电子邮件，对其进行解析，并根据其内容调用不同的api. We deployed it on AWS using SES for notifications, S3 for saving the emails, RDS for saving logs, and Lambda for parsing and calling the API.

该项目专注于使用Python Scrapy抓取深层网络以获取医学信息. I created and fixed crawlers along with the pipeline. The most challenging thing to handle in this project was the scale.

这个项目使管理大学建筑变得更加容易，因为大量的设备不断更新建筑数据. 我创建了一个数据库来存储所有的数据，并在此基础上建立了数据分析机制来处理所有的信息. 我还将所有这些打包到一个Python Dash web应用程序中，并将其部署在Heroku上.

This project analyzed and classified Arabic speech. Working on it was challenging, mainly because I don't speak Arabic. 我必须做一些额外的工作来正确评估不同的模型，并了解哪种模型更好.

I have created many different LinkedIn scrapers during my career. 有些人使用高级搜索功能收集个人和公司数据，并提供最佳的结果过滤. 我还实现了账户轮换，以避免被封禁，并通过在AWS上部署爬虫程序实现了流程自动化.

我创造了这个情绪分析工具来提高股票市场价格预测的准确性. It checks for tweets, analyzes them using Word2Vec and a CNN architecture model, and outputs the sentiment. This project was very challenging!

I created a model for stock market price predictions. Then I added backtesting using a lean engine. Finally, 我设置了一个Interactive Brokers (IB)服务器，并使用IB Python API创建了一个纸质交易(模拟)环境.

I created a scraper for badminton live scores, matches, draws, and players, and I built an API on top of that. The whole project was scheduled and deployed on Amazon EC2. This was one of the biggest scraping projects I've worked on.

This was a huge project for reinsurance companies. I added full reinsurance functionalities in C#. 最困难的挑战是理解复杂的保险逻辑和创建可重用代码.

In this project, 我实现了一个BitTorrent协议，允许用户创建自己的torrent客户端. 这是我使用Python并应用我的网络知识的第一个项目之一.

This was an HTTP server implementation written in C. 在这个项目中工作是具有挑战性和有趣的，因为我必须从头开始实现几乎所有的东西, which expanded my technology experience.