Jun 2021 – Dec 2024 · 3 yrs 6 mos

• Leading 50+ people Personalisation team based out in IDC divided across BackEnd Engineering, Data Engineering & Data Science, overseeing products like Personalised Smart Substitutions AI driven Walmart Grocery substitution program, BuyBox Product which uses AI/ML to select the best seller’s offer for an individual item personalised at a customer level, Unified Badging Platform powering badges in Ecom using ML to increase the sales on ecommerce by helping customers in the decision-making process, Personalised Semantic Search Reranking which uses GenAI and vector search to help 140 million weekly customer finding items they are looking for in Walmart Ecommerce & Retail platform.
• Responsibilities:
• Design the architectural blueprints for designing products and prepare technical documents/papers about them. Define long term vision and roadmap. Mentor senior ICs. Analyse system performance with NFRs. Provide timely feedback and seek and utilise received feedback.
• Guide team on coding standards, guidelines, design and architectural patterns, languages and frameworks with evolving trends in the Industry with proper implementation of telemetry features. Drive automation in CI/CD.
• Anticipate risks in requirements phase and recommend corrective steps. Conduct comprehensive analysis of requirements to create long term solutions. Drive creation of MVP, and then iteratively build in phases.
• Build relationships with stakeholders by articulating complex information consisely and address ambiguity. Build consensus in sensitive situations, present to wide audience positioning arguments and data with well structured story. Become trusted advisor to diverse stakeholders
• Engaging with the relevant stakeholders to develop a coherent view of program success metrics.
• Responsible for managing application security and vulnerability assessment by integrating security scans, Legal policies compliance like CCPA, GDPR

Jul 2019 – Jun 2021 · 1 yr 11 mos

• Worked on Behavioural Targeting(Display Advertising): This platform is used to audiences for Ad Campaigns based on user data.
• Worked on Sponsored Products DS pipelines used for showing relevant product ads to the users searching and browsing on the Walmart website.
• Lead multiple ML-based initiatives like loyalty, propensity with a team of 5 from inception to production.
• Designed reusable, modular, and optimised systems/pipelines built on top of Spark to generate features from Terabytes of user data for Display and Sponsored business.
• Designed modular modelling pipelines, using Spark ML framework, consisting of several stages ensuring parallel processing and intermediate progress(in case of failures).
• Scaling the pipeline capable of running several hundred ML segments per day.
• Built frameworks on top of Spark, Hive and Airflow for the DS team to reduce engineering efforts to build PoCs.
• Evaluated new frameworks and upgraded our systems from time to time.
• Mentoring juniors, providing oversight and help teams solve complex issues in big data domain balancing factors like development time, execution time and cloud cost.
• Functional:
• Experienced in doing product research, placements, competitor analysis, feasibility analysis and then designing, developing and deploying a working system i.e.
• Experienced in capacity planning for each initiative by creating technical tasks out of final product backlogs and then breaking down the tasks into granular details for bi-weekly sprints and then socialize the plan to the team, product and leadership.
• Project progress tracking using Jira and smartsheets(gnatt chart view) and providing weekly progress to the leadership.
• Managing several stakeholder teams(UI, Catalog, QA) at the program level with clear communication and separation of responsibilities with/without the help of Program Managers.
• Actively involved in hiring activities like profile screening, standardising hiring process for ML Engineering, taking interviews.

Sep 2016 – Jul 2019 · 2 yrs 10 mos · Bengaluru Area, India

• Leonardo IoT Platform represent a network of physical objects known as things that are used to collect and exchange data. It helps to set up a thing model, companies, and service personnel in a company and their relationships. Technically it comprises of data pipelines, polyglot persistence stores where timeseries data is ingested, validated, transformed to canonical Thing model and stored. Once the data in inside the system, partners can build domain specific apps like Fleet Management, Predictive Maintenance (using deep learning algorithms) on exposed rest services.
• Designing multitenant big data ingestion pipeline capable of ingesting millions of messages per spark streaming micro batch. Kafka acts as the source of the spark job while Cassandra(warm store), S3(cold store) and SAP HANA (aggregate and metadata store) act as the sinks. Some features of the pipeline are :
• Zero data loss powered by at least once kafka-spark integration and automatic deduplication by idempotent Cassandra cluster, Metadata caching in spark streaming state, removal of skewed data identified by ingestion pattern, smooth recovery of spark job crash using optimised periodic checkpointing, Zero downtime of job during periodic maintenance of spark cluster by enabling High Availability in Spark cluster, Customised monitoring.
• Designing the event pipeline which internally uses SAP Datahub’s flow and an embedded event runtime engine to generate complex events and derived timeseries data based on associated rules on top of the metadata of source device.
• Designed architectures on Data Science problems like Trip Clustering and Geospatial Analytics, Trip Similarity, Trip Classification, Predictive Maintenance for the demo partner apps. These partner apps are microservices developed in spring boot with spring security for authentication and authorisation.
• Other activities include mentoring, choosing technologies, setting development standards, code reviews, writing architecture documents.

Jul 2015 – Sep 2016 · 1 yr 2 mos · Bengaluru Area, India

• Re-engineered/Re-Architected the NLP Pipeline including Tokenisation, POS Tagging, Entity Recognition, Correlation Analysis & Sentiment Analysis to ingest millions of news articles using Java and ClearNLP.
• Used Lucene later elasticsearch for performance, resiliency and robustness, to index 0.5 million documents per hour.
• Implemented Article Summarisation, Finding Popular Topics, News Trends, Statistics, Entity Search (Full Text Search) on the News Articles exposed a restful microservices.
• Optimised/Performance Tuned the Graph based Java implementation of TextRank to find relevant articles using JProfiler.
• Used swagger for api documentations of microservices
• Used JUnit, Mockito amd PowerMockito to develop extensive test cases with 90 percent code coverage.
• Used Jenkins for build automation.

Jul 2014 – Jul 2015 · 1 yr · Bengaluru Area, India

• Designed a Multi Agent Based framework in Java on top of JADE(Java Agent Development Framework). This framework dynamically load codes in statistical programming languages like R. These codes process various machine learning algorithms on the data provided from the Java layer and provides their output back to the Java agent. The whole process is conducted by agents and a whole topology is built on it to achieve the final goal. This design is made to attain selective scalability of various components of these algorithms(Trading Stock Prediction, Correlation between various stocks by streaming data from TWS, Text Processing & Sentiment Analysis on Twitter data).
• Experienced in developing scheduling components using Quartz scheduler.
• Experienced in working with high performant web-sockets library socket.io.
• Benchmarked Message Queues like RabbitMQ and Kafka, by performing a comparative analysis on streaming video data/test data using MQ's with SSL vs w/o SSL.
• Used JProfiler to performance tune applications for both time and memory footprint.
• Experienced in Redis for Caching and In Memory Storage and Cassandra for timeseries data.

Sep 2011 – Jun 2014 · 2 yrs 9 mos · Hyderabad Area, India

• Product: Oracle Payments
• Oracle Payments is a complete payment and receipt processing solution that enables efficient, reliable and secure financial transactions. It is the central payment engine for Oracle E-Business Suite that lowers costs and improves control by integrating out-of-the-box with major processors and financial institutions
• Involved in development and enhancement of the product (Functional and Performance Enhancements)
• Worked on the Crypography related aspects of payment instruments and payment systems
• Technical skills involved are Core Java (for internal processing), Oracle Applications Framework (UI), SQL, PLSQL (core Logic), and UNIX.

Nov 2009 – Sep 2011 · 1 yr 10 mos · Hyderabad Area, India

• Project 1: Nokia Siemens Networks Smart
• Developed multithreaded Batch Processing system which provision 4 lacks of SIM within 24 hours.
• Extensive use of Java thread pooling, database connection pooling, load balancing in the server, optimizing the CPU utilization, optimizing a huge flow of I/O in the system, Secure Network File Transfers using SFTP
• Project 2: WebEx Saudi Telecom Company
• Development and enhancement of a system which converts, verifies the request for making Teleconference by establishing a handshake between CISCO system and TELCO Interactive Voice Response server.
• Technologies used in this were Core Java, JMS, VoiceXML’s, MySQL sever scripts.