Feb 2024 – Present · 2 yrs 1 mo

• Ads Externalisation - Drived and executed the charter on Externalising Ads API's with various 3rd party agencies like ANS commerce, Seller app, Hector, accenture,etc.. Built a generic platform for externalising Ads api's. Making it easy, seamless to onboard and maintain without compromising on the security aspects. Ensuring the advertisers also have a control over their ad accounts through policy managements. Enough checks and balances to keep the system controlled at the same time open it up for agencies. This is expected to give a significant boost to Ads revenue in the coming days to Flipkart
• Drived BBD 2024 for Ads Data platform with various NFR's for all kinds of systems like Data proc clusters, Clickhouse, Druid, Elastic search, Map reduce jobs, Spark jobs, mysql, Reporting services, etc.. Took care of various aspects of resiliency, disaster recovery, performance tuning, cleaning up, cost optimisations to keep the cluster size minimal, etc...
• Drived the charter of anomaly detection platform for Ads Data platform which can detect anomaly on the business metrics early and this can help reduce Ads burn, improve ads revenue,etc.. This is extensible for other systems within flipkart to be onboarded.
• Drived the track of supporting AB analysis for Ads metrics.
• Reviewed and contributed to other solutions like Migration to IceBerge from Pubsub, Fixed system stability issues like zookeeper connection timing out due to GC issues, etc..

Jul 2021 – Feb 2025 · 3 yrs 7 mos

• Had worked on the blueprint of Grocery for the Search systems.
• Contributed to the high level solution and review of some of the high impact projects like List based shopping, Unified experience (Surfacing listings of Flipkart market place in Grocery market place seamlessly to solve selection gap), Instant search, Grocery insights- Generic debugging tool to debug cross cutting Grocery issues eg: to detect buyability & serviceability issues and does first level of triaging and point to the right sub system. This helps in reducing the MTTD and MTTR, Restock widget in Inception page,etc.., Carving out and building a new Past purchase system, KVI reranking (personalised ranking)
• Taking important decisions for long term view of Search like should store trees of BGM and Grocery be unified.
• Built a long term view of making systems multi tenant to onboard&maintain Grocery, Shopsy flows seamlessly along with flipkart flows (with proper isolation at different levels).
• Built a long term view of using product attributes for different levels of ranking within Search(Feature store)
• Tech optimisations:
• Bootstrap time optimisation for the core Search data store.

Jul 2021 – Jul 2023 · 2 yrs

• Retail outbound fulfilment:
• Maintenance of existing legacy services in Retail outbound fulfilment.
• Rearchitect of our systems from monolith to micro-services. RAAS(Retail as a service) - Externalising retail services.
• Leading the team to ensure the hygiene, tech metrics like coverage, component testing are all on track. Migration of all our application to K8s. Managing the scrum and enabling the team members to deliver and unblock any production issues for RAAS to go live.
• Worked on the V2 version of the CQRS system that we built earlier, we added DLQ (Dead letter queue) for this CQRS.
• Simplified Query Handler to support a new client to onboard just through configs instead of writing code by coming up with different levels of abstraction.
• Log Search tool - Have written a command line tool to search logs from different sources like K8s pods, Hive(where logs get ingested). so that the developers don't have to go to different sources to fetch logs. This is similar to stern with additional feature to retrieve logs from Flipkart ecosystem. This improved developer productivity.
• Grocery Search:
• I moved to Grocery team which is a cross functional team that works with different teams in Flipkart. I mainly worked for Flipkart Search as part of Grocery team.
• Identifying the root cause of critical issues in production and fixing them across different components in Search/dependent teams for Grocery. eg: Bug in Dropwizard metrics-core library that caused applications to hang, timeouts in product page to make listings unavailable.
• Worked on some critical features to scale Grocery. Search being user facing team, had a good hands on with optimisations, NFR aspects like throughput, latencies, analysing CPU/MEM/DISK/Network utilisations and identifying right no of pods to scale, scaling, deep dive into internals of JVM and tuning JVM, identifying issues with dependent teams systems and sorting them out.

Jul 2018 – Jun 2021 · 2 yrs 11 mos

• Leading the Private Labels tech team and then retail procurement outbound scrum.
• Private Labels:
• Built couple of products - Insight extractor, Iris(Product life cycle management tool) from the scratch in Private Labels and contributed a series of MVPs like targeted marketing, Fashion intelligence, Listing Management service.
• Part of Tech Five pillar project - to drive alerts and monitoring charter for the SSE.
• In Retail outbound procurement - we were maintaining the legacy services as well as building new services at the same time. Moving from monolith to microservices architecture. Consignment management system, Retail capacity manager, IWIT (Inter warehouse and inventory transfer - legacy service), Titan- Stream processor & Palantir- Query handler ( CQRS based system built from scratch), Maintaining other services like azkaban, DSS(Document signer service)
• In this role, I was lucky enough to have explored many new areas. I have mentored and enabled many engineers, contributed in hiring, gathering product requirements, taking Architectural decisions, design choices, choosing tech stacks within the scrum boundary, sometimes beyond scrum, managing the tech hygiene of the services owned by the team, giving directions to the junior members and unblocking them whenever required, assigning tasks to the individuals, playing the role of scrum master, OKR planning, Workshops, certifications, driving UTs and component testing for legacy systems and new ones,Migrating to Kubernetes and so on.

Jul 2017 – Jul 2018 · 1 yr

• 𝐈𝐧𝐬𝐢𝐠𝐡𝐭 𝐄𝐱𝐭𝐫𝐚𝐜𝐭𝐢𝐨𝐧 𝐅𝐫𝐨𝐦 𝐓𝐞𝐱𝐭 𝐚𝐬 𝐚 𝐏𝐥𝐚𝐭𝐟𝐨𝐫𝐦 𝐒𝐞𝐫𝐯𝐢𝐜𝐞:
• We use data extracted from Customer Reviews, Seller comments, Returns Comments to derive key insights like what are the key aspects to concentrate on while building differentiated products for Indian customers.
• We went through existing research works on extracting insights from text using various NLP techniques like dependency parsing, deep learning based approaches, etc.. experimented with these approaches and converged on a approach based on the data.
• Tech Stack: Python, Django, Pyspark
• Used: NLP techniques like dependency parsing, pos tagging, pruning,etc.. using libraries like Spacy, Nltk, etc.. Machine learning algorithms like Random Forest for classifying opinions, Deep learning techniques like Word2Vec word embeddings for clustering, Fast Text for sentiment classification, etc..
• Contributed in: Going through existing works, doing quick POC's, Architecture, HLD, LLD, development, testing, deployment, scaling and reducing latencies using pyspark.
• Award: ICE Value award for being the core Contributor for this Product and unblocking other team members.
• 𝐓𝐚𝐫𝐠𝐞𝐭𝐞𝐝 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠:
• System to retarget Flipkart customers on social networking sites like Facebook and GDN.
• This system targets customers who are a potential buyer for a particular product based on their past browse history, sales data, etc..
• Contributed in: Defining requirements, collaborating with other teams in Flipkart to reuse existing work,
• Solutioning and doing a POC for the same.
• Responsibilities included mentoring a SDE 1.
• Work is in Progress.
• 𝐏𝐫𝐨𝐝𝐮𝐜𝐭 𝐒𝐩𝐞𝐜𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐏𝐫𝐨𝐩𝐨𝐬𝐞𝐫:
• System to come up with the specifications of the right product to manufacture.
• Work In Progress. We are currently doing a POC for this.

Oct 2016 – Jun 2017 · 8 mos

• 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐭𝐢𝐜𝐬 𝐓𝐨𝐨𝐥:
• Built a simple tool with dashboard showing the rating graph, search rank graph, RPI graph for the Billion(Private label of Flipkart) products in Flipkart and send alerts based on the rules configured and hence enable our Business team to quickly fix any issues.
• 𝐓𝐫𝐚𝐜𝐤 𝐚𝐧𝐝 𝐓𝐫𝐚𝐜𝐞 𝐟𝐨𝐫 𝐅𝐮𝐥𝐟𝐢𝐥𝐦𝐞𝐧𝐭:
• (1) Stores and retrieves all fulfilment related data
• (2) Checking if actions like address change is possible, can item be cancelled, etc... based on rules.
• (3) Real time reports on Fulfilment data.(Not implemented but present in design)
• Tech stack: HBase, Elasticsearch, Java
• Contributed in: Requirements Gathering, System Design, HLD, LLD, Evaluating data stores, Coding, e2e Deployment, Functional Testing, Non Functional Testing, Setting up alerts, Maintenance.
• Solved for transactions in HBase for a specific use case.
• Supported two version for this service, first version directly points to Orchestrator's Fulfilment data store.

Jan 2016 – Sep 2016 · 8 mos

• 𝐋𝐞𝐚𝐝𝐢𝐧𝐠 𝐃𝐂 𝐌𝐢𝐠𝐫𝐚𝐭𝐢𝐨𝐧 & 𝐏𝐀𝐀𝐒 𝐈𝐧𝐢𝐭𝐢𝐚𝐭𝐢𝐯𝐞𝐬 𝐟𝐨𝐫 𝐚𝐥𝐥 𝐭𝐡𝐞 𝐒𝐞𝐫𝐯𝐢𝐜𝐞𝐬 𝐢𝐧 𝐅𝐮𝐥𝐟𝐢𝐥𝐦𝐞𝐧𝐭:
• Flipkart Migrated its services to the new in-house Data centre. I was the representative for Fulfilment team(size: 16) to understand different constructs of new Data centre and Platform as a Service Constructs like Elastic Load Balancer, Cosmos, Configuration as a service, Cloud constructs like instances, instanceGroup, RepoServices, Alert Service, Deployment Service, etc.. and do a quick POC for all of these. Setting up data store in the new data centre and comparing the configurations with the machines in the older data centre with the new data centre.
• Contributed in: Resource estimations, Guiding and unblocking Team members in migrating their services to the new data centre, Devising strategies for migration as in what order the services has to move to new Data centre, migrating Data stores to the new Data centre, etc.. Writing scripts for common problems. eg: generic instance scripts. Contributed a library to the Config Service team to automatically populate config values using Reflections without modifying the rest of the code.
• 𝐒𝐞𝐫𝐯𝐢𝐜𝐞 𝐏𝐫𝐨𝐦𝐢𝐬𝐞 𝐄𝐧𝐠𝐢𝐧𝐞:
• System which promises with a vendor, time slot for after-Sale Services like installation, demo, uninstallation, technician visit to the customers based on the available capacity at that point.
• Tech Stack: Java, Redis
• Built this from the scratch
• Contributed in: Requirements Gathering, Evaluating Data stores, Extensible Architecture, HLD, LLD, Coding, Continuous integration pipelines.
• Award: Good Team Player, Mentor and for quicker delivery of Service Promise Engine
• 𝐒𝐞𝐫𝐯𝐢𝐜𝐞𝐬 𝐅𝐮𝐥𝐟𝐢𝐥𝐦𝐞𝐧𝐭 𝐌𝐚𝐢𝐧𝐭𝐞𝐧𝐚𝐧𝐜𝐞:
• Supporting existing features, built new features like Dropship, returns services fulfillment, Automation and Reducing on-call load.
• Mentored a Developer & a support engineer.
• Tech Stack: Java, Mysql, Hazelcast, Drools, Active JDBC

Jul 2014 – Dec 2015 · 1 yr 5 mos

• 𝐒𝐜𝐡𝐞𝐝𝐮𝐥𝐞𝐫 𝐚𝐬 𝐚 𝐬𝐞𝐫𝐯𝐢𝐜𝐞(𝐀𝐧𝐮𝐬𝐮𝐜𝐡𝐚𝐤):
• Anusuchak is a Scheduler as a service implementation with a REST interface. The communication from anusuchak to the end point is using restbus(Flipkart's internal queuing system similar to SQS).
• Contributed in: Design discussions, supporting new features, testing.
• Tech stack: Drop Wizard, Java, Quartz
• 𝐉𝐞𝐞𝐯𝐞𝐬 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧:
• Jeeves is a Service Partner for Flipkart. We did a deeper integration of Services Fulfilment system with Jeeves systems.
• Tech Stack: Java, Mysql
• Contributed in: Requirements Gathering, Contracts definition, Solutioning, LLD, Coding, Functional Testing
• Award: Quicker and Smooth Delivery of Jeeves Integration.
• 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐧𝐠 𝐃𝐚𝐭𝐚𝐒𝐭𝐨𝐫𝐞 𝐟𝐨𝐫 𝐈𝐝𝐞𝐦𝐩𝐨𝐭𝐞𝐧𝐜𝐲 𝐅𝐢𝐥𝐭𝐞𝐫 𝐢𝐧 𝐑𝐞𝐬𝐭𝐁𝐮𝐬:
• Restbus is Flipkart's internal queuing service that makes it easy to decouple and scale microservices. we evaluated various datastores and benchmarked the metrics like latencies, throughput for the Restbus's access patterns.
• Contributed in: evaluating HBase.
• 𝐒𝐞𝐫𝐯𝐢𝐜𝐞𝐬 𝐅𝐮𝐥𝐟𝐢𝐥𝐦𝐞𝐧𝐭:
• Orchestrator for fulfilling after sales services. CP system which listens to events from upstream systems and creates, tracks and orchestrates service item entities and pushes events to downstream systems.
• Tech stack: Java, Mysql, Dropwizard, Drools, Hazelcast
• Contributed in: Coding, Functional Testing
• 𝐒𝐞𝐫𝐯𝐢𝐜𝐞 𝐃𝐞𝐬𝐤:
• Service desk is a ticketing tool for tracking after sales service tickets.
• Building new features and supporting/optimising existing features like download, paginated search, etc.. in Service Desk Tool.
• Tech Stack: Ruby, Padrino, Mysql
• Contributed in: Coding, Functional Testing