Data Engineering with focus on building large-scale data pipelines.

AI for Marketing Compliance: Recovery of 20K conversions/day under EU DMA act using a hybrid XGBoost model;

designed and deployed system to maximize ad reach on unconsented data

Data Driven Attribution: Delivered accurate campaign ROI measurement for advertisers by leading the end-to-end

development of a multi-touch attribution pipeline; from project inception through design and rollout, built with SQL, Airflow, and MapReduce to capture and integrate full-funnel B2B buyer journeys

Cross collaboration with Product team and Data Science team to drive the results

Conversion APIs: Led the Attribution effort which increased revenue by $20M (30% lift) despite third-party cookie deprecation through a secure offline-to-online attribution API; engineered using Scala, Spark, Hadoop, Azkaban

Leadership in Data Privacy: Established privacy-first AI practices under GDPR, ATT, and DMA by partnering with

Signals and AI teams; ensured fairness, governance, and robustness using data annotation and monitoring

Hypothesized chaos experiments to increased system resiliency, designed fault injection and failover strategies

Working on development of AI solutions to enhance lumen virtual edge compute platform by forecasting failovers for predictive maintenance in edge infrastructure

Ideated synthetic data generation methodology leveraging chaos engineering on triple O openstack hyper converged infrastructure to address lack of available data for prototyping production environment for AI, shipped synthetically generated logs to Zabbix and ELK stack using filebeat and metricbeat modules

Produced SWOT analysis, low level design document, derivation of problem statements, execution plan

Designed and built virtual network infrastructure automated by zero touch provisioning on FreeBSD 12.1 and Juniper platforms for team's internal purpose. Within set up, with ansible remotely added security, VPN, virtual firewalls, BGP routing policies

Improved spinning up Centos 7/8, Ubuntu 18.04/20.04 and FreeBSD 12.1 Qemu-kvm VMs by creating golden images with additional capabilities specifically for Lumen Tech to standardize security and save time for package installations on new VMs

Front end development using Sencha Architect and Oracle database.

Predicting ship’s position in stormy zones is critical for the shipping companies. Applied various Machine Learning techniques to predict the position e.g. random forest, K-means clustering, gaussian process to get the best results. Weather data to determine stormy regions and oceanographic data to determine wave heights and ocean currents, are used to draw the stormy regions. The locations of the weather and oceanographic data are matched with the the grids of the ships’ locations and timestamp to optimize the algorithm. Later ML algorithms were implemented in a distributed environment using Dask to study the behaviour of the ships in/around stormy region to predict the safe location away from the storm for the ships, which are presently heading toward them, making sure they are not deviated much from their expected path. Technology: Python, Dask

Subject - Starting Computing