JUST BE YOURSELF, THERE IS NO ONE BETTER
YOUR REPUTATION IS OFTEN YOUR RESUME
Design & Analysis of Algorithm, Database Systems, Foundations of Artificial Intelligence, Machine Leaning for Data Science
Design and Analysis of Algorithm, Data Structures, Software Engineering, Database Management System, System Software, Operating Systems, Computer Organization, High Performance Computing, Machine Learning, Topics in Deep Learning, Computer Networks, Modeling and Simulation, Optimization, Complex Networks, Discrete Mathematics, Calculus
Rank: 5 (out of more than 300)
Secured 99.89 percentile in grade 12
Rank: 70 (out of more than 70000)
Working with Compiler team at Sigma, I added functions’ support for data warehouses like Redshift, Snowflake, Google BigQuery, MySQL, Postgres, and Trino. I also worked on adding JSON datatype natively within Sigma to support BigQuery Json-input functions, impacting an increase of 20% in adoption rate and a decrease in latency by 30%.
As a CP for this course, I oversaw course operations, including assisting students with 350+ questions, grading assignments and exams, and creating exam solutions for selected questions.
Working with the Amazon Pay Later team, I implemented an end-to-end manual suppression module to handle third-party planned downtimes by suppressing the Amazon Pay Later payment method on PSP. Then, I extended it into automated suppression by continuously monitoring ASR (Attempt Success Rate) and saved 3-6 hours of manual work per week for on-call. I also worked on many ad-hoc tasks such as infrastructure setup & management, operating system migration, and upgrading deprecated packages. Finally, I created the "Amazon Pay Wealth" product as one of the lead software developers, indulging in designing, infrastructure creation, end-to-end coding, testing, and release for over 150 million customers with efficient scalability and fault tolerance.
We devised a robust method to automate the process of finding erroneous data points in the crowdsourced eBird dataset using outliers determined by the DBSCAN clustering algorithm, adjusted to boost robustness and minimize false negatives in the cluster. Then, we assigned a reliability index to each observer based on identified noisy points to remove unreliable observers from the analysis. This algorithm proved to be very potent and could be easily generalized, so the study is in process of publication in a reputed journal called Avian Research under the name Flagging Malicious Contributors in Citizen Science Datasets.
I built an end-to-end KYC 1:1 video calling KYC application that worked even with low data connectivity using a technology called WebRTC based on peer-to-peer communication. I also built a Kurento Server for a server-side video recording mechanism to protect customers’ privacy and avoid possible data leaks. I also created a base Android Application using WebView for rendering this application's web pages inside the application.
SKILL IS BETTER THAN STRENGTH
ACHIEVEMENT IS MANUFACTURED BY HARD WORK
EVERY PROJECT HAS CHALLENGES, AND EVERY PROJECT HAS ITS REWARDS