SOWRYA GALI
+1-929-***-**** ad2rkt@r.postjobfree.com LinkedIn: sowrya-gali GitHub: sowgali
EDUCATION
Columbia University
Master of Science, Computer Science
New York, NY
GPA 3.8/4.0
Courses Cloud Computing, Distributed Systems, NLP, Causal Inference Activities Qualified and participated in ICPC GNYR 2023 Indian Institute of Technology, Hyderabad
Bachelor of Technology (Hons.), Computer Science and Engineering Hyderabad, IN
GPA 9.11/10
Awards Excellence in Research
Courses Database Systems, Parallel and Concurrent Programming, OS, Network Security EXPERIENCE
Columbia University Prof. Junfeng’s Lab
Graduate Research Assistant
New York, NY
Sep 2023 - Present
● Leveraging diffusion models to remove noise for test-time adaptation improving object detection and robustness. (Generative AI, Diffusion Models).
● Designed and developed codebases for diffusion models and object detection pipelines in a distributed setup for Cityscape’s Dataset using Detectron 2 and guided diffusion. (PyTorch, Distributed Training) Kelly Services Data Science
Software Developer
Remote, US
May 2023 – Present
● Architected and implemented revenue forecast pipeline for various businesses of Kelly with continual training as data arrives leveraging MLFlow. (Databricks, Spark)
● Designed and developed a novel ensembling technique that combines STL Forecasts, and ARIMA and built an in-house automated ARIMA to improve the shortcomings of existing libraries that aid in ensembling. (StatsForecast, Pandas, Python)
● Accelerated the existing forecasting pipeline by combining LGBM and Autoregressive models and integrating them with Spark.
(Apache Spark, Databricks, Scikit-Learn, Time Series Analysis, MLFlow)
● Accurate forecasts integrated into PowerBI aided in making faster business decisions with precise budget allocations. Gruppay Backend Core Services
Founder
New York, NY
Jan 2023 – Present
● Architected and developed a microservices-based backend for Gruppay to handle complete and split payments integrated within the chat. (Stripe, Java, StreamChat)
● Designed payment handling engine to ensure correctness and high throughput under high-stress concurrency conditions. (Apache Kafka, GCP, Spring, Redis)
● Managed and overviewed the app development cycles and will soon be available to the public after rigorous testing. (Docker, Kubernetes)
Columbia University Prof. Misra
Graduate Research Assistant
New York, NY
Jan 2023 – May 2023
● Deployed honeypots at various locations of Columbia’s Network including inside and outside DMZs to monitor potential attacks
[over 10K interactions/day] on the network. (T-Pot, Linux, Pandas)
● Analyzed the collected data and designed various metrics to measure the severity of the attacks and attackers for generating telematics for cyber insurance. (ElasticSearch, Python)
● This work is presented as a poster in Columbia’s Data Science Day, attended by delegates from top companies. BNY Mellon Collateral and Clearance Technology
Software Developer
Chennai, India
May 2021 – Jul 2021
● Prototyped and formulated an application to track transactions and clearances over global markets with plugins to customize desired data to be displayed. (Angular, Spring, SQL)
● Spearheaded front-end development to let users select markets as map pointers using a search bar with auto-complete features and integrated it with Google Maps API. (TypeScript, Angular)
● Programmed microservices to monitor requests from users and data updates [over 50K updates per day] in the backend leveraging Spring and managed project timelines efficiently and delivered one week before the deadline. (Java, Kafka) SKILLS
Languages C/C++, Python, Java, JavaScript (React/Angular/Typescript), MySQL, Go Technologies AWS, Docker, Kubernetes, Spark, Redis, Elasticsearch, MongoDB, Spring ML PyTorch, TensorFlow, Scikit-Learn, HuggingFace, Pandas, Statsmodels, CUDA PUBLICATIONS
● Anindya Sarkar, Sowrya Gali, Vineeth N Balasubramanian, Empowering a Robust Model with Stable and Object-Aligned Explanations, AROW-ECCV 2022
● Anindya Sarkar, Anirban Sarkar, Sowrya Gali, Vineeth N Balasubramanian, Get Fooled for the Right Reason: Improving Adversarial Robustness through a Teacher-guided Curriculum Learning Approach, NeurIPS 2021
indicates an equal contribution