Principle Data Scientist & Data Engineer

Data Science/Engineering/Architecture:
* Co-designed a new Real Time Data Streaming Architecture using Kafka + Apache NIFI
for on premise data acquisition and Google Pub/Sub, BigQuery and DataFlow for public
analytics pipeline.
* Migrated on-premise Big Data pipelines to Google Cloud using Airflow and DataFlow and
Apache NIFI .

Software and tool development:
* Designed CI/CD pipeline workflows for in house developed apps using internal Gitlab
runners.

AI / Machine Learning / Mathematical Optimization Focused Projects
* Optimized Sky NowTV , SkyGO and SkyQ , Netflix and Google Youtube traffic mapping to
subscribers and saved considerable costs. Used certain heuristics to find the optimum
path and server placement.
* Using Fast Fourier transform and supervised machine learning (AdaBoost) to train
traffic models for CDNs and classify different problems based on changes in traffic
patterns. This automated ML pipeline has reduced mean time to respond.
* Using unsupervised learning techniques identify city level video habits.

Software and tool development:
* Built Python Django and Flask based web apps for network analytics and automation,
saving thousands of engineering hours since 2015.
* Developed REST APIs and web apps for Machine Learning models .
* Open sourced PyGoBGP and PyGNS libraries for wider network automation community
working on the same problems

Statistics and Data Analysis Focused Projects
* Using statistical bootstrapping ran A/B testing for NowTV and SkyGo UK video quality
streams .
* Designed statistical experiments considering Statistical Power Analysis and measure
network path impact on video quality for the video content served to Germany from UK.
Results influenced CTO level decisions for infrastructure investments in Germany
* Quantified performance of internet exit paths using Internet scale measurement
experiments. This was a unique project with some academic depth and findings were
shared with wider internet community in public talks.