Jul 2023 – Dec 2024 · 1 yr 5 mos · San Francisco Bay Area · Hybrid

• Working on the AI Perception team:
• Training and fine tuning of UNets, Latent Diffusion Models, DinoV2 for perception
• Using foundation models for perception
• Deep teaching and generative simulation for augmented data
• Optimizing on device performance

May 2022 – Jul 2023 · 1 yr 2 mos

• ML for Healthcare @ Google Research now DeepMind
• Built algorithms to understand the medical condition of a patient based on EHR data
• Training ML models to understand tabular data and large language models for doctors’ notes
• Worked on health insights and model stability improvements with Med PaLM

Nov 2021 – Apr 2022 · 5 mos

• Worked on building a context aware representation for care-provider and patients’ interaction
• Representation includes health insurance cards and previous medical history of the patient

Mar 2021 – Oct 2021 · 7 mos

• Worked on time-series based anomaly detection models for account compromise detection
• Using out-of-distribution detection on composite features built upon users’ email profiles
• Used weak supervision to decrease model maintenance as distribution drift occurs

Jul 2020 – Jan 2021 · 6 mos · Mountain View, California, United States

• Working on ultra low power ML. Now IYO
• Fixed term position till Jan 21

May 2019 – Aug 2019 · 3 mos · Seattle

• Alexa AI

Jan 2018 – May 2018 · 4 mos · Greater Delhi Area

• Teaching Assistant for the course of Probability and Stochastic Processes
• Took tutorial sessions for problem-solving and further reading

Jul 2017 – Dec 2017 · 5 mos · Greater Delhi Area

• Teaching Assing for the course: Communication Engineering
• Developed weekly problem sets and took tutorial sessions for problem-solving and further reading

May 2017 – Jul 2017 · 2 mos · Bengaluru Area, India

• Worked on designing, developing and deploying an auto-moderation system for
• book campaigns
• Designed a text based model to produce feature vectors for the given campaign
• from its custom text and description
• Developed an end-to-end training and testing pipeline for weekly training builds
• and live scoring of incoming campaigns
• Deployed this model to production for batch-level scoring of a set of incoming
• campaigns

May 2016 – Jul 2016 · 2 mos · Bengaluru, Karnataka, India

• Worked on handling undefined op-codes for an architectural simulator
• Developed a layer to handle instruction level access for the CPU and the execution
• of exception return
• Compared native performance with the simulator and improved upon the perfper-watt
• characteristics. Used QEMU to emulate an ARM environment for CPU
• architectural testing

Nov 2015 – Jan 2016 · 2 mos · Gurugram, Haryana, India

• Worked on Machine Learning algorithms to optimize image search and object detection for an
• e-commerce website. It included working on CNN Architectures to extract features from images and then classifying those images via SVMs and Bag of SIFT words classifier.
• Used an integration of ROS(RobotOS) and Caffe to build the object detection pipeline with the given classifiers.