Leveraged technical expertise to design and develop REST APIs, providing robust solutions to complex technical problems for a user base of over 50K active users.
Collaborating with product managers and cross-functional teams to comprehend business requirements, leading feature design, and producing comprehensive documentation.
Identifying, debugging, and resolving production issues, ensuring system stability through proactive monitoring and fine-tuning.
Conducting thorough code reviews, providing constructive feedback to enhance code quality and maintain best practices.
Took ownership of identifying and addressing security vulnerabilities, resulting in vulnerability-free code. Recognized with the Biased Towards Action award for my contributions at PubMatic.
Conducted a proof-of-concept on File Uploading to Slack Channel via Code + Load Data command in MySQL, resulting in optimized DB-intensive workflows and improving the Viewability Calculation flow by 60%.
Improved monitoring and alerting processes, resulting in a 25% increase in detecting P1 and P2 issues and improved platform stability.
Secured 4th place in the PubMatic hackathon - 2021 for developing the Inventory Quality Dashboard, a tool that calculates a transparency score to verify the authenticity of the parties engaged in programmatic advertising transactions.
Mentored 6 new hires, providing guidance and support during their onboarding process, and conducted interviews for 20+ candidates, contributing to the hiring of top talent for the company.

Java Spring Boot MySQL Apache Kafka Microservices Docker Kubernetes Kibana Graphana CI/CD Jenkins

Conducted experiments with SST-2 and TREC-6 datasets to study how text classification models (CNN, LSTM, and BERT) rely on important keywords. Found that these models tend to prioritize keywords over understanding the meaning of sentences, which can limit their classification accuracy.
Developed an innovative approach called ShufText to uncover the limitations of these models and their excessive reliance on important keywords.
Applied mathematical, statistical, and machine learning expertise to design and execute experiments, focusing on utilizing BERT.
Published a research paper titled ShufText: A Simple Black Box Approach to Evaluate the Fragility of Text Classification Models at the LOD'21 International Conference.

Python 3 Keras BERT CNN LSTM Jupyter notebook

Implemented a SaaS-based subscription management software to facilitate the company's product offerings through subscription models.
Collaborated with UX designers and Machine Learning engineers to develop captivating data visualization dashboards that empower users to easily comprehend critical information and make informed decisions.
Worked with a variety of different languages, platforms and frameworks such as Angular 4, TypeScript, Javascript, Flask API, Google Cloud Engine.

Angular 4 Flask HTML CSS JavaScript Python 3 GCP Plot.ly