Dec 2017 – Jul 2019 · 1 yr 7 mos · Gurugram, Haryana, India

• Employee Engagement Platform for Grant Thronton: an ML-based platform that makes informed, data-backed decisions on hiring, promotion, and resource allocation; created a Chatbot using CoreNLP, and ElasticSearch which predicts a conversation's sentiment and maps the context to an organizational driver; classified chat questions for cold-start; created a feedback loop to enforce
• learning. Used Spark, ElasticSearch, JAX-RS, RESTful, MySQL.
• Exercise tracking system: a real time exercise repetitions counting and person recognition system that uses Deep Learning to count the number of repetitions of approximately the same exercise in an live input video stream and simultaneously recognise and track the person by face. Used CNN,Python, Dlib, OpenCV.
• Personalization Platform for Voot Kids: created an ML-based recommendation engine to power cross-platform widgets on Voot Kids App; Used Spark, Redis, JAX-RS, MySQL, and ElasticSearch.[QPS<1000/s; P95<50ms].
• Dispatcher System for Nagpur Metro: moved the traditional desktop application to a SOA; created a real-time peer-to-peer voice streaming application using WebSockets, EDA, WebRTC on Kurento.
• Smart Health Coach for CureFit: created a food nutrition catalog from external sources by crawling, cleaning, and deduplication; developed a rule engine which created personalized health plans for users using ElasticSearch, MySQL, and JAX-RS. [QPS<500/s; P95<20ms].

Jul 2015 – Dec 2017 · 2 yrs 5 mos · Gurugram, Haryana, India

• UCMS (Unified Customer Messaging System): Built an event-based, highly scalable centralized communication pipeline on a distributed storm cluster for sending out all transactional communications to customers. The complete processing of an event is done in ~400ms by selecting templates, fetching data and sending communications.
• Technologies used: Apache Storm, Kafka, MongoDB, MySQL, Aerospike, Java, Spring [QPS<800/s]

Jun 2014 – Jul 2014 · 1 mo · Nodia,India

• Released a new version for the company's flagship Android Application "Mark Attendance". Incorporated support for new APIs in attendance stats and redesign the user interface
• using Material Design.