Joined the team that focuses on scalability and production deployment in the search system vertical1. Introduced cell-based architecture for production deployment and management of search systems, with following high level and key deliverables: a. Designed and developed a new serverless control plane b. Redesigned data plane workloads for Kubernetes compatibility, simplifying how search catalogs are deployed and handle traffic c. Leveraged vertical scaling to make workloads stateless, streamlining autoscaling d. Implemented vertical scaling in the existing VM architecture, achieving a 10% reduction in cloud costs e. Dockerized the search application, saving an additional 20% on cloud costs by testing feasibility on AWS Graviton f. Bootstrapped the project end-to-end, from infrastructure to domain-driven design, facilitating team onboarding 2. Experimented and delivered a brand new hybrid search design that leveraged vector search + lucene and LLM for offline data cleaning and imputing. Elements of design later productionized by core search team as Search+ 3. Deep dove into Apache Solr and Apache Lucene to resolved key migration bugs transitioning from Apache Solr 6 to Solr 9, addressing issues in bootstrapping time and cache misses

In very simple terms AWS Quicksight Q is natural language search engine for data visualization1. Initial core member since early steps of the product2. Setup initial software onboarding guidelines and code architecture3. On ML side created the first E2E benchmark and helped in setting up data collection iteratively for model improvement4. Designed and architected unique distributed systems problems and races to our data plane5. Helped ramp up many devs to AWS standards, backend engineering and other important technology choices and learnings at AWS

As part of this role I had to design and build systems following AWS standards i.e reliable, secure, scalable, fault tolerant. Also these systems altogether were designed to present ML based solution to customers. A lot of dev-ops writing cloudformation scripts and since I was also part of a team that had deep ML knowledge, got to work on some ML cool problems as well.Details of my contributions:1. Designing NoSql/Dynamo based architecture keeping in mind concurrency and races in distributed systems, versioning etc2. Working with tech lead to help him come up with architecture and request flows to avoid race conditions3. Owning significant chunks of OO design & implementation for dynamo models, backend workers, test canaries4. Writing lot of cloudformation that represents our test infra and also setting up internal tool for team for rapid development and iteration5. ML wise wrote some spark related code that utilized ML as well as successfully debugged an ML issue of seeing unexpected Precision/Recall numbers. Overall ML wise gained a lot of intuition and background of traditional supervised ML and challenges that come with it.

A big change in my working style after coming from a startup. I learnt more formal methods of communicating, zeroing in on customer impact since any minor change had the possibility to cause significant damage to revenue and DevOps because of the famous "Amazon oncall" process.Major learnings:1. Service Oriented Architecture2. Ownership in a distributed environment My successful deliverables have been:1. Developed a micro service from scratch. This involved disambiguating business and technical requirements, working with multiple teams inside and outside Amazon, creating service architecture, getting approval from internal security and open source teams and designing and developing the service2. Helped in launching Amazon Households3. Bringing in better coding practices in the team e.g using Lombok, better code reviews, pursue refactoring etc4. Setup integration tests for Households which included - fighting through a legacy testing framework, creating a software design for modelling data driven testing and setting up a CD pipeline5. Created a mini MVC framework on top of our legacy web framework to help us write better code with controller and views unit testing6. Five months worked from India while delivering results, improving my remote working skills drastically

Working at TakeOff has been the most challenging and satisfying experience so far. Building up a startup, putting in money, breaking expectations, meeting expectations and myriad of different situations and pressures accelerated my growth big time.SKILLS that I learned1. Unexpectedly picked up appreciable design skills. Designed a shop’s interior, few flyers, one seminar event, a website, few logos, a billboard and a newspaper ad in a span of 5 months2. Advanced web development. Developed entire frontend of shopify based e-Commerce shops3. Expertise in Frontend technologies - Less, Sass, Grunt, Gulp, AngularJS, Bootstrap4. Developed an ERP backend in Ruby on Rails and ajaxified it through angularjs5. Developed a CRM using Google Apps and Google script6. Deployment engineering using DockerQUALITIES which I gained1. Being productive working from home, after 1 and a half year experience2. Collaborating and getting stuff done from consultants3. Developing based on customer behavior and experience 4. Team hiring and mentoring

Rovi constantly tries to innovate on the experience of media consumption on TVs. Snapstick was acquired by Rovi because of the alignment of vision. So from Snapstick, where I was involved in developing a product similar to chromecast, I shifted to enhancing TV guide experience by taking it to mobile. It was a pretty great experience. I started my career from there, learning a whole wide spectrum of software development encompassing servers, deployments, javascript and mobile apps. Few projects that I have worked upon:1. Collecting real time sensor data from mobile for gesture prediction. This was used to create a tennis game which you can play on TV using your cellphone as a racket2. Worked with game development companies - Sarbakan & Playjam, to integrate Snapstick product into HTML5 and Unity based games3. Created windows installer of Snapstick using NSIS4. Created a Facebook open graph crawler in python for video recommendation engine5. Created a module to switch application to larger screen automatically on both Mac & Windows6. Developed a HTML5 multiplayer game from scratch, working in close co-ordination with overseas game designer7. Crucial role in porting Rovi ios app to android

1. Developed javascript based data structure which brought down searching 10k records from 10sec to 1sec2. Hands on experience with Microsoft technologies - SSRS, SSAS & ASP.NET