Akhil Surapuram

Graduate Research Assistantship under Dr. Sudhir Aggarwal in ECIT Lab developing a forensics intelligence platform (SM-TDES) that would be able to selectively extract data from multiple phones and provide valuable AI/ML insightsto law enforcement agencies during mass incident events.

• Leading a geographically distributed team of 5 Research Assistants (RAs) across 3 universities, including 3 from UAH (University of Alabama in Huntsville) and 1 from Purdue University, while collaborating with two other professors on this multidisciplinary research project.

• Conducted extensive experiments to fine-tune and evaluate face detection libraries, leveraging the insights gained to develop a Python flask microservice capable of accurate face detection and recognition on a given set of images.

• Designed and developed high performance geospatial API for loading cases into android device, enabling dynamic case assignment based on the device's geographical location.

• Ownership of investigator panel Vue.js frontend, revamped the entire web App, resulting in improved user experience, and streamlined data accessibility. Implemented advanced features in "Photos Page", enabling investigators to apply multiple filters (e.g., locations, time, face, objects etc.) with "and & or" conditions for enhanced query flexibility. https://youtu.be/UflzN2gIj18

• Implemented various data science analytical features like photos filters & timeline. Worked/Experimented on a diverse range of models, including deep learning for generative AI, image captioning, natural language processing (NLP), reinforcement learning and distributed ML.

As Data Engineer – built payments ETL pipeline for extraction of payments metrics (payments dataset) into warehouse for powering dashboards for data analytics and enabling data-driven decision-making for our business teams.

• Achieved a remarkable 99.1% reduction in the execution time of a legacy data pipeline, optimizing performance and resolving bottleneck issues, resulting in a significant improvement from 8 hours to just 4 minutes.

• Successfully migrated for over 35+ pipelines, seamlessly incorporating older pipelines into a new warehouse setup, ensuring smooth data flow and uninterrupted analytics processes.

• Increased data warehouse platform reliability close to 100% by fixing edge cases in data ingest and Egress modules which has resulted in data inconsistency.

• Created a data sync client for Kaleidofin Credit Analytic services “Ki score” – An Enterprise Integration tool which is capable of automated data syncing irrespective of schema & database type, Achieving high speed, low latency, heavy data transfers between Ki score partner and kaleidofin.

• Guided external Data Engineering vendors and Made Architecture related changesfor supporting single Tenancy and Multi Tenancy of Data warehouse architecture.

• Mentored my co-workers and sub-ordinates, ensuring best practices through code reviews, pair programming and design reviews of various modules related to backend server and Data Engineering Stack

• Made changes in CB’s core FW for accepting Non-Deterministic API patterns. Like allowing path variables containing slash (‘/’).

• Day Based Billing Project, API Rate limiting Project.

• Performed query performance analysis and optimized long running queries