Oct 2019 – Present · 6 yrs 5 mos

• Search Systems Backend:
• Single handed design, coded core Search APIs from REST to gRPC in Search core Systems as well as Search middle-wares keeping backward compatibility in mind. Impl Automated test suit for Search core systems and Search middle-wares, verfiied against live sample user queries. Moved Search to gRPC end to end with feature
• gate with no downtime and errors. Overall perf gain: significant perf improvement in response times & bandwidth.
• Undesrtood the need of resiliency across all Search systems, executed POC for Circuit breaker and Ratelimiter for following libs -> resilience4j, Alibaba’s Sentinel, Guava RL, Netflix Zuul. Design and implemented CB and RL for all Search Systems.
• Implemented Apache Storm metrics collector
• Updated legacy Apache Solr schema, which involves identifying unused vars, optimizing Solr index, updating schema and Index in Solr DB without downtime. Significant gain in RTs & Bandwidth.

Mar 2017 – Oct 2019 · 2 yrs 7 mos

• Search & Recommendations backend Systems:
• Looks Recommendation Tier 2 Platform (Lisa) :Any combination of styles/product which can be worn together is called looks ( Tshirt, Jeans, Shoes, Sunglasses,Cap .
• Cross sell: Recommendations contain other recommended products which could be paired with the purchased product to create a look.
• Shoppable Looks: Allow users to shop products (same or similar) that the model is wearing in PDP.
• core platform:
• Delivery time based filter and sort = >
• Initial design and Implementation based on Solr Inner Join .Delivery based filter enhancement from inner join to Solr block Join. Design & implementation of block join Indexing pipeline .
• Indexing Pipeline for products data:
• > Real time processing of push & pull based indexing events using Apache Storm stream processing framework & store in database like Mongo, Solr, Redis.
• > Real Time Indexing will take the indexing calls from clients where real time update is required like inventory, discount and product details. Design & implement push & pull model.
• Mobiles Apps:
• Worked on various third party libraries like UIL, Fresco, FB Connection-class, FB Network Bandwidth , Volley , RoboSpice , Retrofit and customised as per requirement .
• Mobile Apps:
• > UIL customization {width & height as input & restrict duplicate downloads in memory, return width & height specific image to optimise memory}
• > Fresco customization {expose method to synchronous check for image present in disk & men cache}
• FB Connection-class customization {new nw bandwidth range for slow nw in India & download image as per nw, device, bucketization}
• > Developed various Utilities like MYNSchedular {offload work non UI & UI work efficiently anywhere within app}

Sep 2014 – Mar 2017 · 2 yrs 6 mos

• Myntra Mobile Apps:
• > Designed & developed customised Expandable ListView to handle variety of diff layouts eg { Horizontal & Vertical Lists, Non expandable list, expandable list, split banners list }
• > Offline Push notification system }.
• > MynACo is an Analytics library which can be used = > { integrate multiple Analytics tools such as Google Analytics, Localytics , Apsalar with an easy to integrate abstraction layer. Library also includes ability to accumulate analytics data off from UI thread & schedule plus dispatch to Analytics tools in background thread }
• > Designed and developed end to end Customer Onboarding System using which app can dynamically present diff screens with different onboarding to customer. System includes key terminologies such as strategically representing Onboarding on the client side and marking its completion on the server side for all customers.
• > MYNCoachmark { utility to show UI onbaording pop ups anywhere in apps using custom views }.
• > Major focus on performance of app and snappy user experience which includes slow netowork performance in India for bandwidth like 15 to 20 kbps , memory , speed.
• > Image Downloading pipeline used in the entire app which also controls diff image size downloads based on network bandwidth on the fly .Diff cache size allocation based on device bucket.