Spearhead the development of innovative products and platforms by translating business requirements into technical solutions.Drive the adoption of AI across the organization, leveraging advanced techniques in fine-tuning and pre-training large language models (LLMs).Guide and mentor aspiring engineers, fostering their growth into future tech leaders.Collaborate with cross-functional teams to align technology strategies with business goals, ensuring seamless execution and delivery.

Building data platform for delivering enriched user mobility dataDesigned & developed audience sets , which enables analysis with large sets of millions of items ; this allows sets to be combined, queried at large scaleDesigned & developed 'splice machine' a framework to extract data & metrics across various dimensions. This allows for faster , smarter data extracts for analysis.Developed summarisers for generating summaries for different dimensions(location, app & combined). Evolved with a concept of 'UserPresence' which forms the basic abstraction for deriving insights.Developed an algorithm to model user presence & infer activityHave lead the development 2nd generation of dataplatform, have actively managed this from Sep2017 to Dec2018.Developed highly scaleable data pipe to ingest , process & sink RTB data streams at high velocity & volumeDesigned a hierarchical multi stage rule engine which process upwards 10K rules on 4B+ bid dataDesigned & developed a data-warehouse to allow tera-bytes to be ingested and queried with Spark & Parquet

Developing scalable systems to allow users to use api dataDeveloped a query processor atop apache spark(SparkSQL) to allow data to be queried from S3. query processor manages adhoc queries in highly multi tenanted deployement. Built an intellitgent routing method for queries for different clusters, this allowed to build a predictable system.Developed a framework to analyze, score & alert for anomalies in Api transaction data, this allows users to quickly idenitfy, correlate failuresDeveloped an generic aggregator framework on Spark framework, this allowed existing system to scale up, details http://bit.ly/1WsruOfin Apigee, learnt Using Apache Spark to build frameworks to crunch dataBuilding systems in Microservice architectureA bit about cloud operations & aws

My role as technical architect involved designing new architecture for processing large data sets & developing solutions to analyze application behaviourDeveloped analytics module for databases which tracks slow , misbehaving SQL queries & detects anomalous database states; this could detect anomalous DB states & SQL queries with less than 2% FP rateDeveloped anomaly detection module to detect anomalies in server(Web/App) behavior & application transaction volumes; The anomaly detection was quick within 3s of data arrival & with accuracy of 95%Applied clustering & regression methods to analyze load-behavior variation of systems & to predict behavior of systems under different load conditionsDeveloped a solution for predicting hotspot methods in a java application using decision trees(C5 & RandomForest)Designed & Developed REST based service for metrics & configuration entities for external & internal consumersDesigned & Developed services & DAO layer for serving metric & configuration extraction from a multi-tenant DBWorked extensively with Polyglot data sources(MySQL & HBase)Prototyped data-pipeline for ingesting & processing application monitoring data which could process at the rate of 20K tps; The architecture derives much from the Lambda Architecture for ingesting, processing & reporting on large data sets; Apache Storm is used for real time stream processing & Apache HBase is used for data storage & reporting.Prototyped event stream processing atop of Apache Storm for real time queries on application performance data; this involved using sketch algorithms to efficiently process streams for counting & cardinality estimationKey SkillsRegression, Decision Trees, Ensemble methods,ClusteringApache Storm,KafkaWorking knowledge of Hadoop & MR Task Composition

NNMi Smart Plug-in for Quality Assurance for networks(8.X & 9.X series)Worked in translating the requirements, design & development of the product to monitor the quality of service innetworks.Lead the product development of management of the Class Based QoS(CBQoS) in networksDesigned and developed the solution for the management of probes (IPSLA) to monitor network performanceDesigned & developed the domain-model using JPA/HibernateWas part of the team which developed multi tenancy across the productsCompletely owned several modules like domain-model, configuration, discovering network topology, thresholdmonitoringPrototyped to analyze Call Detail records & Netflow Data using the Map Reduce paradigm with HadoopDeveloped prototypes to validate the usage of Cassandra to be used as time series database for performancemetricsDeveloped prototypes to adopt the anomaly detection algorithms for the network performance metrics

Ranking anomalies in Data centersExtensively worked with scientists in implementing and validating anomaly detection and ranking algorithmsDeveloped prototypes to aid in adoption of the new techniques in outlier detection and rankingWork on statistical outlier detection & exploratory data analysis of time series (metrics in data centers)

Prototyped and validated many aspects of the new architecture adopted in the new productsDesigned & developed the domain-model using JPA/Hibernate, also worked on other products helping in designingthe domain-modelWorked on network polling & monitoring and root cause analyzers, developed root cause analyzers for analyzingL3/L2 VPN faults and service impactsWorked on Jboss as App Server, JPA/Hibernate to develop the domain model

Parallel Discovery of Network TopologyInvolved in developing modules to support the parallel architectureDeveloped many tests to measure the performance of the new parallel discovery paradigm

Worked in fixing many complex issues and technical escalations• Developed new features to enhance the product• Had helped the Field engineer in deploying the product in different customer environments