May 2019 – Jul 2019 · 2 mos

• Formulated E2E design for ML pipeline to solve generic speech-related tasks. Developed deep residual network to detect Disfluency in different types of speech.

Feb 2019 – May 2019 · 3 mos

• Accelerating AI through Distributed Environment: Developed Resnet architecture based ML Model using ISIC 2017 dataset. The model is trained under the IBM cloud based distributed environment. Learned the process of training the model based on 12app factor methodology and compared the performance with vertically scalable GPU environment.

Feb 2019 – May 2019 · 3 mos

• Worked on venture capital funds dataset performing several crucial tasks.
• 1. Developed visualisation pipeline to be used by other team members.
• 2. Developed a preprocessing and systematic cleaning pipeline for all the extracted raw data.
• 3. Normalised all the sectors and filled other missing values in all the datasets.
• 4. Visualised North America VC dataset using different visualisation techniques.
• 5. Removed redundant questions in the list of questions which need to be answered using visualisations.

Aug 2018 – Jun 2019 · 10 mos

• Conducted algorithm and data structures classes for competitive programmers at both intermediate and advanced level. Covered topics such as graphs, dynamic programming, trees, segment trees, fenwick trees etc. Encouraged competitive programming at all levels. Also conducted many programming contests at college level as part of technical fest on popular coding platforms such as hackerearth.

Jun 2018 – Jul 2018 · 1 mo · Bangalore, India

• Responsible for major design decisions for migrating Amazon Pay documentation to Amazon's common portal. Designed framework for making retry calls in async auth workflow after the failure of sync authorization.

Dec 2017 – May 2018 · 5 mos · Gurugram, Haryana, India

• Implemented a Learning Management System on Web and Android which uses LDA tag detection and smart neural network to score enrolled students in provided course and published on Google Play Store.
• Features Include:
• 1. Several types of tests are provided to calculate the student's profile score
• 2. Text recognition in Android app to scan student's marksheet
• 3. Topic Modelling to generate tags accoding to the student's skills
• 4. Hexaco Personality Test
• 5. Courses and Projects with quizzes
• 6. Each course, project and test contribute to a student's score via a dynamically trained neural network