Jun 2023 – Oct 2024 · 1 yr 4 mos · Chicago, Illinois, United States · Hybrid

• Improved the performance of the existing price prediction systems, by employing techniques such as integral closed- form solutions and Numba JIT compilation, reducing computation time from 20 seconds to less than 1 second.
• Led the development of a new Python-based price prediction system, experimenting with different algorithms and expanding coverage across 8 exchanges.
• Streamlined the trading data visualization service, enabling the addition of new data through simple configuration changes instead of cumbersome code modifications.
• Led the development of an automated Python-based borrowing service, allowing traders to view, request, and repay loans across multiple exchanges using a single platform.
• Skills: Python, SQL, Databricks, Solace, Docker, Jenkins, Ansible, Nginx Proxy, DNS, Prometheus, Grafana, Loki, Numba, Asyncio, Bokeh

Jan 2023 – May 2023 · 4 mos · Los Angeles, California, United States

• Teaching Assistant under Professor Mohammad Reza Rajati for CSCI 360 - Introduction to Artificial Intelligence

Aug 2022 – Jan 2023 · 5 mos · Los Angeles, California, United States

• Teaching Assistant under Professor David Kempe for CSCI 270 - Introduction to Algorithms and Theory of Computing

Jan 2022 – Aug 2022 · 7 mos · Los Angeles, California, United States

• Course Producer under Professor Shahriar Shamsian for CSCI 570 - Analysis of Algorithms

Jan 2023 – May 2023 · 4 mos · San Mateo, California, United States

• Designed a Java and SQL-based solution to revamp the interface between the billing and growth teams, resolving various issues associated with the Snowflake trial accounts.
• Created a Jenkins build pipeline to streamline the team’s release process, saving approximately 2 hours of engineering time with each release.
• Skills: Java, SQL, Snowflake, Shell Scripting, Jenkins, Docker, Git, IntelliJ Idea

Jun 2022 – Aug 2022 · 2 mos · Miami, Florida, United States

• Developed a machine learning-powered solution to predict the likelihood of an application not meeting its Service Level Agreement (SLA) in the next 30 days.
• Experimented with various ML models, such as the SARIMA time series prediction model, to improve prediction accuracy from 35% to 80%.
• Streamlined the application onboarding process by automating the hyperparameter search for the model, saving roughly one week of engineering time typically required to identify a good set of hyperparameters for each app.
• Skills: Python, Statsmodels, Matplotlib, Terraform, Kubernetes, EC2, AWS Lambda, Jenkins, Grafana, Prometheus

Jul 2020 – Jul 2021 · 1 yr

• Improved the accuracy of a URL-based ad-targeting algorithm by using NLP techniques such as embeddings, TF-IDF, and clustering, leading to a significant increase in annual revenue from under $200K to over $3MM.
• Optimized the team’s data layer, by leveraging efficient data serialization formats, like Avro, reducing startup time from 45 minutes to 15 minutes, thus facilitating faster iterations and testing for engineers.
• Reduced the 99th percentile response time of a large-scale distributed topic similarity service, from over 50 seconds to 2 seconds. Achieved by implementing various optimization techniques, including database-level caching, dataframe indexes, and join optimizations.
• Skills: Python, Java, SQL, Flask, Gunicorn, Tomcat, Docker, Jenkins, Avro, Redis, Grafana, Prometheus, Kibana Elastic Search, Tensorflow, Keras, CUDA, Pandas, Numpy, NTLK

Jan 2020 – Jul 2020 · 6 mos

• Boosted the accuracy of a webpage brand-safety classification service from 55% to 90% by employing natural lan- guage processing techniques such as BERT embeddings & the LSTM recurrent neural network model.
• Designed an efficient Java-based, multi-threaded crawler capable of crawling and scraping the entire Wikipedia in less than 24 hours.
• Skills: Python, Java, SQL, Flask, Gunicorn, Tensorflow, Keras, HuggingFace, Pandas, Numpy, NLTK, Scikit-learn

May 2019 – Jul 2019 · 2 mos · Bangalore

• Designed a Java and Python based data archival system to automate the transfer of data from a Sybase IQ Database to AWS S3 buckets, resulting in savings of more than 99% in storage costs.
• Achieved a 60% size reduction by using the Parquet data format along with GZip compression.
• Skills: Python, Java, SQL, SybaseIQ, AWS S3, Parquet, GZip, Git, IntelliJ Idea