MLE at Amazon General Intelligence org building Amazon Nova models

Full stack developer that owned critical service- Owned service responsible for app installation, app management, app migration and app recommendation, app request features on atlassian marketplace and embedded marketplace. Part of oncall roaster• Ideated an app recommendation system using association rule mining technique in Atlassian embedded marketplace. Developed the first backend prototype in Kotlin that constitutes 3.5% of app installs• Enhanced and owned the core installation legacy service to identify and resolve license generation software defects which reduce app installation failure by 35% approximately saving around 36 Million USD in a year 2021.• Optimized load management and batch emailing pipeline to launch request app feature on the embedded marketplace which led to 1.2M USD revenue in the following quarter• Updated MongoDB data model and exposed GraphQL API for the Cloud Fortified App program - to provide end customers a set of extensively secured top 10% of cloud appsTech Stack Used: Java, Kotlin, DynamoDB, React, Javascript, Git, CI/CD, Spring, SpringBoot, React, JQuery, RESTful API,Ansible, CI/CD pipeline, Git, Linux, Agile, Scrum, Docker, DynamoDB, MongoDB, Amazon Web Services, MySQL, Node.js, Microservices, Maven, Gradle, JUnit, Enzyme, Test-driven development (TDD), Redux, Selenium, Cypress, Mockito, Tensorflow, PyTorch, OpenCV, Matplotlib Numpy, Pandas

Redesigned UI filters on the marketplace homepage by developing reusable React components in Javascript which led to 4.3% increased filter usage• Created a Kotlin microservice and designed DynamoDB schema to build scalable distributed storage, and query systems that are fault-tolerant and easy to manage to facilitate migration of on-prem apps to the cloud with Rest APIs• Secured marketplace against privilege escalations and XSS attacks. Implemented the SLOs as API latency and availability of EC2 machines on SignalFX(terraform), providing real-time SLO violation alertsTech Stack Used: Java, Kotlin, DynamoDB, React, Javascript, Git, CI/CD, Spring, SpringBoot, React, JQuery, RESTful API,Ansible, CI/CD pipeline, Git, Linux, Agile, Scrum, Docker, DynamoDB, MongoDB, Amazon Web Services, MySQL, Node.js, Microservices, Maven, Gradle, JUnit, Enzyme, Test-driven development (TDD), Redux, Selenium, Cypress, Mockito, Tensorflow, PyTorch, OpenCV, Matplotlib Numpy, Pandas

Engineered features and build and trained ML models using machine learning algorithms around the novel problem of detecting mental health problem based on user’s mobile phone app usage. I also constructed end to end data pipelines using Amazon Web Services • Explored the analysis of mobile usage data as both supervised and unsupervised problem with unreliable labels. Engineered features with the domain experts and performed data analysis on the novel dataset• Compute mental health score based on mobile phone usage and built machine learning models in python that diagnosed depression with the recall rate of 75% and precision of 71% • Developed a real-time pipeline using Amazon Web Services Sagemaker to calculate mental health score • optimized a pipeline that ran for 24 hrs to 5 seconds • Built data pipeline using AWS services (Glue, Sagemaker, DynamoDB, API Gateway) fetching data from mobile application to train machine learning model

Teaching Assistant for Data Structure and Algorithms.

We worked on developing innovative diagnostic tools to address unmet clinical challenges through integrating biology/medicine, micro- and nanotechnology, consumer electronics, and artificial intelligence.Successfully designed Convolutional Neural Network models that classify virus with the accuracy of 95% and able to distinguish between sample and control with the accuracy of 97%Analyzed Hepatitis B Virus genome sequence using LSTM to predict its resistance towards 5 most common drugsEnumerate Neutrophil based on their morphology using Computer Vision techniques for cancer and infection detection

Setting up the core for the chapter1) InductionACM inducted 13 new members for the upcoming year. The student were selected based on their coding skills and general computer science aptitude.2) Alumni NetworkWe connected with the Computer Science Alumni who were willing to connect with ACM and its activities.3) ACM ConstitutionWe with the help of Abhishek Thakur sir made a constitution detailing on following aspects-PURPOSE, VISION AND MODUS OPERANDI, Core List of ACM Student Chapter ActivitiesJob Description of office bearers, Financial Budget and Management of funds PreambleOrganizing events for publicizing ACM Student Chapter1) Neural Network Workshop SeriesACM with the help of a few seniors organized Neural Network workshop series. The participation was high throughout the series. Following classes were organized to help students-IntroductionANN - Theory and ImplementationCNN - TheoryCNN ImplementationRNN / LSTM TheoryRNN / LSTM Implementation2) Computer Science Department ProjectWe connected with the Computer Science Department who were willing to discuss the project and work they are currently engaged in.Faculties were given slots where they discussed their projects, current progress and how students can contribute to the projectThey also informed the student about several future possibilities in their respective field

Project: The aim was to come up with a Maintenance Schedule for fleet management companies that improve the overall reliability and availability of the system while simultaneously maintaining the budgetary constraint.Genetic Algorithms: Since the problem was comprised of multiple objectives and their nonlinear structures, we decided to solve the model using multiple objective meta-heuristic Genetic Algorithms(3 Fitness Function) Results: Successfully completed the model that presented and analyzed a new mathematical function that models age reduction and improvement factor parameter for repairable and maintainable components