- Implemented bulk inferencing at scale with Kafka, FastAPI, Kubernetes Jobs API- Instrumental in developing and implementing metrics-driven model building framework- Spearheaded color detection research and boosted color palette prediction by Semantic Segmentation in Studio and Wild images- Collaborated with Product Attribute API engineers and accelerated model building by crafting an end-to-end model build/deploy pipeline- Conceptualized and established a fully serverless infrastructure for ML inference workloads in collaboration with MLops, ML Engineers and Data Scientists across the team- Mentored MLops engineers in AWS cross-account migration with zero downtime- Organised cross-team collaboration between ML Engineers & Data Team for VogueAI

- Evaluated and implemented serverless (zero cost at idle) deep learning model deployment framework- Managed cloud account migration with zero downtime- Singlehandedly built/maintained 5+ Kubernetes clusters on AWS- Reduced infrastructure costs during the COVID-19 pandemic (50% MoM reduction)- Managed 50+ Tensorflow/Pytorch models at scale- Built fault-tolerant, auto-healing CI pipelines for deploying to Kubernetes - Ensured zero downtime via CI deploy checks (automated rollbacks on broken code deploy)- Architected workflow pipelines and integrating them with in-house products- Optimized allocated resources across applications- Built Infrastructure management APIsManagerial Responsibilities included- Occasionally putting out fires- Ensuring the growth of reportees; making them feel safe and comfortable, ensured mental wellbeing during the pandemic- Cross-team collaboration- Onboarding new team members and making them feel welcomeProducts with significant contributions have been linked below

On paper role: Security Engineer -- Data ScienceWorked as a part of the Innovation and Research Team - Researched and created ML-based unsupervised time series anomaly detection - Designed a novel technique for ranking security alerts for prioritizing actionable items based on the severity of the alert. - Designed, developed and deployed a data modeling schema for correlating event fields across multiple data sources in ELK. - Created a full stack flask-based tool for easy manipulation of ELK data pipelines.

Asset Management System - Designed and implemented the full stack using HTML, CSS, flask. - Developed a chat assistant using Dialogflow which helped users in identifying products and prices while placing PO requests.Fleet Optimisation - Developed visualizations using Google Maps APIs using simulated location data. - Captured RFID card swipes using a Raspberry PI and updated passenger count through web sockets.

- Designed the UX and UI for Aanganwadi app using HTML/CSS and Cordova. - Application designed to run in low bandwidth conditions in rural areas of Rajasthan, with 100,000+ kids as users. - Developed content delivery logic for delivering Audio, Video, and PDF books. - Worked in collaboration with one team member, for the backend.

MooID: Image biometrics based Cattle insurance App - Implemented and deployed REST API based image classification using Flask, TensorFlow, Python. - Implemented pattern matching and image segmentation for biometric verification using OpenCV, Python.

Tasks included-Managing Network Infrastructure,Server Administration,Workstation Administration,User Rights management,Website Designing,Services Integration,Hardware Troubleshooting,Software Troubleshooting,Network Troubleshooting