Dragos-Mihai M.

✔ The client, world’s largest online retailer, was missing out on potential revenue from advertising due to the analysis tool for customer behaviour lacking important metrics as the data needed was vast, continually changing and stored into an inaccessible database. • 𝗢𝗽𝗲𝗻𝗲𝗱 𝘁𝗵𝗲 𝘁𝗼𝗼𝗹 𝘁𝗼 𝟭𝟬𝟬’𝘀 𝗼𝗳 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿𝘀 by designing an on-demand system capable of onboarding datasets from the legacy application by setting out an implementation plan and communicating closely with the team members. • 𝗗𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝗮 𝟮𝘅 𝗺𝗼𝗿𝗲 𝗰𝗼𝘀𝘁 𝗲𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲 𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻 by building a rock-solid infrastructure that was fully hosted in AWS by using the CDK library to deploy the Step Function workflow that was in charge of moving the data to the team’s database, encrypting it using KMS and notifying the customer via email when the process was over. ⚙️ Skills: Java, NodeJS, AWS Step Functions, AWS ECS, AWS Lambda, API Gateway, DynamoDB, S3, CDK, CloudFormation, IAM, KMS, Angular, Typescript

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, had difficulties launching the beta version of a customer behaviour analytics tool that had the potential to boost the generated revenue through advertising due to the fact that the core data needed to fuel the tool was stored in an external database in a format not suitable for custom queries. • 𝗜𝗺𝗽𝗿𝗼𝘃𝗲𝗱 𝗱𝗮𝘁𝗮 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 𝗰𝗮𝗽𝗮𝗰𝗶𝘁𝘆 𝗳𝗿𝗼𝗺 𝗺𝗲𝗿𝗲 𝗚𝗕𝘀 𝘁𝗼 𝗼𝘃𝗲𝗿 𝟮 𝗧𝗕𝘀 𝗼𝗳 𝗱𝗮𝘁𝗮 by deploying AWS cross-account resources meant to support the spawning and management of EMR jobs in charge with the datasets transfer which were triggered automatically at the beginning of each month by a cron job. • 𝗨𝗻𝗹𝗼𝗰𝗸𝗲𝗱 𝟮𝘅 𝗺𝗼𝗿𝗲 𝗽𝗼𝘁𝗲𝗻𝘁𝗶𝗮𝗹 𝗿𝗲𝘃𝗲𝗻𝘂𝗲 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝗱 𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀 by making the data available in the team’s system, representing a major stone in launching the beta version of the tool, resulting in funding for growing the team. ⚙️ Skills: AWS EMR, EC2, S3, AWS VPC, KMS, IAM

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, was struggling to increase automation when generating banners for advertising campaigns due to the current system lacking the option to let graphic teams input their own generation rules. • 𝗜𝗻𝗰𝗿𝗲𝗮𝘀𝗲𝗱 𝘁𝗵𝗲 𝘀𝘆𝘀𝘁𝗲𝗺’𝘀 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 𝗽𝗲𝗿𝗰𝗲𝗻𝘁𝗮𝗴𝗲 𝗳𝗿𝗼𝗺 𝟮𝟬% 𝘁𝗼 ~𝟱𝟬% by redesigning the priority based logistic to apply on banners the properties with higher-precedence and improving the DynamoDB database to support a global secondary index. • 𝗘𝗻𝘀𝘂𝗿𝗲𝗱 𝟭𝟬𝟬% 𝗰𝗿𝗼𝘀𝘀-𝗯𝗿𝗼𝘄𝘀𝗲𝗿 𝗰𝗼𝗺𝗽𝗮𝘁𝗶𝗯𝗶𝗹𝗶𝘁𝘆 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗯𝗮𝗻𝗻𝗲𝗿𝘀 by in-depth researching css specific properties for the main browsers and updating the unit tests suite to validate the presence of these properties. ⚙️ Skills: Java, AWS Lambda, DynamoDB, S3, CloudFormation, React, Typescript

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, was unable to set a proper roadmap or evaluate the progress, needs & strength areas of their automatic banner generation system due to neglecting to collect any kind of data in the 1 year since launching the system. • 𝗥𝗲𝗱𝘂𝗰𝗲𝗱 𝘁𝗵𝗲 𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗱𝗲𝗯𝘁 𝗯𝘆 𝟴𝟬% by establishing a dashboard with numerous operational and business metrics by tracking all the actions available to the customers in the system and publishing custom data to CloudWatch. • 𝗥𝗲𝗱𝘂𝗰𝗲𝗱 𝘁𝗵𝗲 𝘁𝗶𝗺𝗲 𝘁𝗼 𝘁𝗮𝗸𝗲 𝗮𝗰𝘁𝗶𝗼𝗻 𝗳𝗿𝗼𝗺 𝟲 𝗵𝗼𝘂𝗿𝘀 𝘁𝗼 𝟭𝟱 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 by in-depth researching CloudWatch monitoring features to set precise alarms which were used for the on-call shifts to notify the team whenever there was a critical or potential outage. ⚙️ Skills: Java, CloudWatch, AWS Lambda, DynamoDB, API Gateway

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, was dealing with a high load of requests and repetitive work for the manual graphic team due to missing in the automation tool a transition shape feature that would separate text area from imagery. • 𝗨𝗻𝗯𝗹𝗼𝗰𝗸𝗲𝗱 𝗮𝗻 𝗶𝗺𝗽𝗼𝗿𝘁𝗮𝗻𝘁 𝗰𝗹𝗶𝗲𝗻𝘁 𝗼𝗳 𝘁𝗵𝗲 𝘀𝘆𝘀𝘁𝗲𝗺 𝘄𝗵𝗶𝗰𝗵 𝗹𝗲𝗱 𝘁𝗼 𝗮𝗻 𝗶𝗻𝗰𝗿𝗲𝗮𝘀𝗲 𝗼𝗳 𝟮𝟰% 𝗶𝗻 𝘁𝗵𝗲 𝗮𝗽𝗽𝗿𝗼𝘃𝗮𝗹 𝗿𝗮𝘁𝗲 𝗶𝗻 𝗹𝗲𝘀𝘀 𝘁𝗵𝗮𝗻 𝗮 𝗺𝗼𝗻𝘁𝗵 by designing a custom-tailored algorithm to identify the orientation of the newly built transition shape component into the banners’ layout and defining a React component to render the visual shape representation. • 𝗥𝗲𝗱𝘂𝗰𝗲𝗱 𝘁𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿 𝗼𝗳 𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝘁𝗶𝗰𝗸𝗲𝘁𝘀 𝗼𝗽𝗲𝗻𝗲𝗱 due to bad banners quality 𝗯𝘆 𝟯𝟬% by closely working with the UX Designer to set a baseline of good banners and updating the integration tests to validate the system’s health against the baseline. ⚙️ Skills: Java, AWS Lambda, S3, DynamoDB, API Gateway, CloudWatch, Angular, React, Typescript

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, was facing potential revenue losses ahead of the Prime day event due to scalability problems for their automatic banner Generation tool as the graphic team was expecting a load of requests exceeding their bandwidth and the automation tool was missing critical event specific fields in its workflow. • 𝗜𝗻𝗰𝗿𝗲𝗮𝘀𝗲𝗱 𝘁𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿 𝗼𝗳 𝗮𝗽𝗽𝗿𝗼𝘃𝗲𝗱 𝗯𝗮𝗻𝗻𝗲𝗿𝘀 during the Prime day event 𝗯𝘆 𝗺𝗼𝗿𝗲 𝘁𝗵𝗮𝗻 𝟱𝟬% by redesigning the backend APIs to use updated serializers & deserializers, migrating the database schema to match the new models and integrating the new fields into the AWS Step Function workflow used to orchestrate each step of the banner’s creation. ⚙️ Skills: Java, AWS Step Functions, AWS Lambda, DynamoDB, AWS RDS, S3, API Gateway, CloudWatch, Angular, React, Typescript, NodeJS, Ruby on Rails

Location: Iaşi, Romania

✔ The client, an outsourcing company helping big organizations like UNICEF and World Bank, was looking to build a reliable notification system for emergency situations since most applications on the market didn’t guarantee safe delivery of the messages due to missing a failover system based on multiple channel communication. • 𝗗𝗲𝗹𝗶𝘃𝗲𝗿𝗲𝗱 𝗮𝗻 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝘁𝗵𝗮𝘁 𝘄𝗮𝘀 𝟵𝟬% 𝗿𝗲𝗹𝗶𝗮𝗯𝗹𝗲 𝗶𝗻 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝟯𝟬 𝘀𝗲𝗰𝗼𝗻𝗱𝘀 by designing a 3 channel communication solution that would first try to reach out the customer by sending Firebase push notification to the custom-emergency mobile application and if there was no acknowledgment on this channel it would move on to sending text message and lastly voice call using Twilio. • 𝗦𝗰𝗮𝗹𝗮𝗯𝗹𝗲 𝘁𝗼 𝟭𝟬𝟬’𝘀 𝗼𝗳 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿𝘀 by building a backend system that was in charge of selecting the impacted users from the database and handling the notification & acknowledgment workflow by using RabbitMQ to process the batches of messages, making the application ideal for local emergency events like flooding. ⚙️ Skills: Java, Spring, Hibernate, RabbitMQ, Android, Firebase, Twilio, JQuery, Bootstrap, HTML, CSS

Location: Bucharest, Romania

✔ The client, provider of cloud services, was not able to monetize their hosts auto-scaling service due to missing a system that could evaluate the health of the service by assessing the accuracy of machine learning models which were forecasting metrics such as cpu, memory load etc. • 𝗔𝗰𝗰𝘂𝗿𝗮𝗰𝘆 𝗼𝗳 𝗠𝗟 𝗺𝗼𝗱𝗲𝗹𝘀 𝘄𝗮𝘀 𝗰𝗼𝗺𝗽𝘂𝘁𝗲𝗱 𝟱𝘅 𝗳𝗮𝘀𝘁𝗲𝗿 𝗯𝘆 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗰𝗮𝗹𝗹𝘆 𝗽𝗿𝗼𝗰𝗲𝘀𝘀𝗶𝗻𝗴 𝗚𝗕𝘀 𝗼𝗳 𝗱𝗮𝘁𝗮 𝗶𝗻 ~𝟯𝟬-𝟰𝟱 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 by designing a system of 2 Map-Reduce jobs with custom implementations of input format and secondary sort that was applying Cross Validation procedure to evaluate the machine learning models on a limited data sample. ⚙️ Skills: Java, AWS EMR, S3, R

Location: Iaşi, Romania

✔ The client, world’s largest online retailer, was wasting hours of the team’s bandwidth on manually updating the metadata for one of their analysis sales systems due to usage of a CLI tool that was prone to errors and strenuous when handling complex data structures. • 𝗥𝗲𝗱𝘂𝗰𝗲𝗱 𝘁𝗵𝗲 𝘁𝗶𝗺𝗲 𝘁𝗮𝗸𝗲𝗻 𝗯𝘆 𝘁𝗵𝗲 𝘁𝗲𝗮𝗺 𝘁𝗼 𝗰𝗿𝗲𝗮𝘁𝗲/𝘂𝗽𝗱𝗮𝘁𝗲 𝗺𝗲𝘁𝗮𝗱𝗮𝘁𝗮 𝗳𝗿𝗼𝗺 𝟭𝟬-𝟭𝟱 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 𝘁𝗼 𝗮 𝗺𝗮𝘅𝗶𝗺𝘂𝗺 𝗼𝗳 𝟭-𝟮 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 by setting up a web application to input the complex structures on a user-friendly web page and handling the operations on backend by interacting with other services and processing the structures to match the UI needs as the rendering was taking place on the server side. ⚙️ Skills: Java, SpringMVC, Apache Tomcat, JQuery, Bootstrap, HTML, CSS

Location: Iaşi, Romania