Dec 2021 – Dec 2024 · 3 yrs · Gurugram, Haryana, India

• Signal generation and trading strategies execution infrastructure in low latency C++
• Market data research systems and post-trade analysis tooling in Python
• Limestone is Tower's largest trading team, and the software I work on runs the core of many unique and varied quantitative trading strategies every day globally.

Feb 2021 – Apr 2021 · 2 mos · Bangalore Urban, Karnataka, India

• Worked on Cloud PC project in Applied Science Team.
• Collaborated on Machine Learning models to predict user behaviour leading to cost savings and better user experience.
• Built pipeline to collect, clean and pre-process user activity data in an encrypted format to protect user privacy.
• Implemented and compared various machine learning models and statistical processes for best prediction in real-time with dynamic data.

Jan 2021 – Nov 2021 · 10 mos · San Francisco Bay Area

• Part of team that landed Series-B funding, supported growth and product teams as a backend engineer, leveraging PostgreSQL, Kotlin and Kubernetes
• Shipped rate limiting APIs, integrated Plaid Exchange, Cash Back Rewards, Fees management, transactions revamp and automated account closure
• Built functionality to assist customer service and risk teams use internal software

May 2020 – Aug 2020 · 3 mos · Germany

• Studied Survivable Network design k-Connectivity Augmentation Problem
• Studied online metric algorithms with untrusted predictions for robust systems. Analyzed lower-bound for weighted caching in a streaming environment
• Proposed novel reductions to Steiner Tree for special instances of 2-CAP
• Theoretically analyzed DP algorithms for the Steiner Tree reduction and obtained improvement in time complexity of 2-CAP approx algorithm problem

Nov 2019 – Mar 2020 · 4 mos · Singapore

• Experimented SOTA deep generative model-based anomaly detection algorithms in PyTorch and performed a comparative study
• Collaborated on streaming multi-aspect group anomaly detection project and published a research paper in WSDM 2021
• Proposed and implemented an Auto-encoder with LSTM model for online anomaly detection

May 2019 – Aug 2019 · 3 mos

• Added state-of-the-art symbolic Markov chains, random matrices, random walk and multivariate distributions algorithms in stats module
• Designed API for compound distribution and random variables. Implemented algorithms to simplify expressions from compound to univariate distribution
• Reduced the complexity of exiting sampling algorithm by integrating external lib. and streamlined the API, achieving 20-40% reduction in sampling time
• Included in SymPy developer map (amongst 43 major SymPy contributors)

May 2019 – Jul 2019 · 2 mos · Hyderabad Area, India

• Identified the problem of anomalies in business and server metrics. This involved working with multiple stakeholders and analysing large datasets
• Designed and implemented a real-time and scalable solution for anomaly detection and correlation-based RCA that supports monitoring and alerting
• Integrated and deployed the RCA library to internal services. It is used for finding anomalies in production business metrics with improved accuracy