Created a CRM solution for Flowcode encompassing the following key features:- Engineered a flexible form designer enabling customers to gather contact details seamlessly.- Implemented the use of these forms as landing pages when end users scan QR codes

1. Cost analytics Pipeline: Built an analytics pipeline for analysis of cost attribution for GCP services/components (attribute cloud cost to different dimensions such as tenants and services): - Developed an Apache Beam pipeline to analyze AppEngine logs (from bigquery to bigquery). Deployed to GCP's dataflow. - Designed pipeline using golang to aggregate data on different dimensions. - Wrote Looker reports and dashboards2. Transformer: Wrote a python transformer for "T" of the ETL pipeline to ingest customer product/purchase data: - Defined a json-based DSL to define transformation - Uses pandas to transform data. - Developed canned transformer. - Transfrom from Google Cloud Storage (GCS) to GCS3. Bigquery Data Extractor: developed a micro-service in golang that serves as "E" of the ETL pipeline.

1. Authorship verification: Determine if a set of documents are writing by the same author or not. Developed mostly on AWS. - Used Apache Spark for data science/data engineering: 1. Extract data 2. Transform data 3. Create golden dataset 4. Extract features - Used SparkML libraries to train model: xgboost, knn, logistic regression. - Experimented with Tensorflow for Deep Neural Networks - Experimented with Locality-sensitive hashing (LSH) to analyze cohort data. 2. Mosaic Plagiarism Detection (re-arranged sentences): - Created several golden datasets: wiki, known cases of mosaic plagirism. - Indexed document based on Part-of-Speach(POS) tags: Create a hash of POS with window of 8, 12, 15 etc. (n-gram of POS tags)3. Export Paper Submission data: Export data for both customers and Business Development team. - GDPR compliant - Used Apache Spark to import/anonymize and export data - Data for business development and customers (unizin)4. Event Ingestion: Data warehousing using AWS Redshift - Data pipeline using AWS firehose and Redshift - Wrote the API and Domain Specific Language(DSL) querying data getting5. Wrote a backend micro-service for document annotations.6. Wrote backend micro-services for advanced paper grading7. Worked on Architecture Board to review and submit new requirements

- Designed and implemented cascaded notification process- Designed and implemented push notification - Worked on various notification component (SMS, Phone, Email)- Implemented various processors based on Message queues.

Enterprise Search: - Developed enterprise search using Lucene/Solr - Developed custom analyzers and Filters - Developed multi-lang and multi-tenant support - Developed C++ and Java APIs for the search - Developed C++ processes that ingest data to Solr - Developed a Codec(PostingsFormat) for SOLR/Lucene allows for fields specified in the schema.xml file to be indexed into a separate key-value database, such as Berkeley Database.

1. Architectured and developed complete student analytics infrastructure for processing and displaying student exercise data: - Capture learning events (reading, practice and progress) - Aggregate learning metrics - Wrote reports (hicharts)2. Progress Forecasting(predicts district level gain in reading): - Created forecasting model - Deployed forecasting model - Wrote the UI for reports

1. Clicktrack Streaming: - Wrote clicktrack streaming pipeline including JavaScript and C++ API that ingest events. - Analyzed clicktrack data to: 1. Improve search and recommendation engine 2. Improve landing page 3. Improve suggestions - Wrote analytics reports to: 1. Find defeciency in knowledgebase 2. Find if certain knowledgebase items are useful or not - Improved chatbot search results.2. Data Aging: - Wrote data/link aging algorithm3. Ant Colony Optimization Algorithm for suggested links: - Worked on Ant colony optimization algorithm that recommends links on landing page