Shubham Nimbalkar

Building multi-substrate cloud infrastructure for Gen-AI research and products

Collaborating with IBM to onboard watsonX

Building spark-on-kubernetes on-prem infrastructure to horizontally scale Cell Manufacturing data jobs by 10x

Maintain real-time analytics for 3 factories on the in-house dashboard that provides yields at different granularities

Leveraging best practices with Kafka, Spark, Hudi and Clickhouse to optimise for latency and compute resources

Designed data pipeline for genealogy mapping from a cell to raw materials across different shops in manufacturing

Built long running streaming jobs with E2E tests and easy deployment that can consume millions of messages per hour

Early adoption of internal multi-device frameworks to automate and migrate jobs in current test pipeline

ShieldSense - A dashboard for teams to visualise xcode project coverage metrics to improve software quality

Developed end-to-end Campaign Intelligence Platform to improve ad-based targeting and boost campaign sales by 5%

Designed data pipeline and cloud infra using spark, elastic and SQL to fetch and co-relate 20 GB of data daily

Built a frontend with HTML, CSS for marketing team with scalable performance using Kubernetes and Docker

Analysed empirical data and competitor insights to expand user base and provide targeted incentives

Created LSTM based sales predictor to visualise sales 2 weeks into the future with more than 70% accuracy

Face Liveness Detection for Rakuten Pay - A face authentication project to make secure payments by thwarting 2d and 3d masked attacks. Collected face depth data of 300 people by IR sensor using python to create LFW dataset

Completed a research project under guidance of Research Scientist, Nithish Divakar to describe an image with tags ranked by visual importance. Designed web application to upload images and inference the AI model.

Aesthetic and Visual Analysis - Proposed automated video metadata extraction to OTT platforms (like Viki/Rakuten TV) to reduce human annotation effort by 90%.

Created an AI pipeline to process a video and output good quality, distinct and scene classified posters of movies

Built Item bundle recommendation engine to boost cross-item selling & average basket size using TF-IDF

Analysed merchant’s selling history, tokenized japanese text and created a similarity scoring matrix for catalogue

Developed a hierarchical server health monitoring and alerting system ‘CEAT’ (Similar to APM systems)

Created node-link graph using java to understand server dependency and failure impact

Presented proof of concepts to integrate voice command operations in CRM app with java

Built demos with Lex & Lambda, api.ai (now Dialogflow), CMU Pocketsphinx along with a survey of others

Integrated a trigger word similar to “Ok Google” in the app’s unbound services