Saurabh Jain

-Created training and evaluation pipeline with pytorch for deep learning based motion correction of rs-fMRI performing realignment of all volumes and regress motion noise.

Regulations lead

- Spearheaded recruitment of regulatory associate and consultant for regulatory team to achieve company's objective of being a ISO 13485 certified body within 9 months with no previous regulation experience.

-Developed Quality management Systems by reviewing and implementing over 10 SOP and 25 records for Design and development to ensure efficient SDLC of the software.

- Represented the company in ISO 13485 and manufacturing license audit , successfully achieving the certification with minor non-conformances.

- created end- to-end modular quality assurance and quality management system to flag discrepancies in incoming clinical data designed to support other MRI modalities and option to be deployed in cloud or on-prem systems.

- Worked with team to migrate microservices into monolithic framework implementing state machine architecture to reduce infrastructure cost by 40%.

Voxelbox - SMRI

- Architected robust ,scalable and disease agnostic structural MRI processing pipeline optimized to process multi modal MRI data using vectorization and multi processing improving the runtime by 50%.

- Developed highly maintainable software , scalable APIs and containerized AI power services in Azure Cloud platform using python and fastapi.

- Build , fine tuned and deployed multiple deep learning models to solve for skull stripping and image denoising.

Voxelbox-tumor segmentation

- Created , containerized and deployed multi modal tumor segmentation model using RESUNET architecture ,working with MONAI(pytorch based framework) for dataprocessing and NVIDIA DALI to achieve 88% dice score for intra tumor segmentation.

- Improved data economy of product by successfully training the model with 3 structural MRI modalities and achieving similar results as SOTA models , leading to better product viability.

- led and supervised in-house MRI data annotation team to create unique testing dataset verified by expert radiologist for validation of the model.

Developed end to end AI backend pipeline for covid
detection and severity quantification from data
collection, data pre-processing ,curating DICOM
dataset to model training using CT data. Developed
and applied different architectures of Convolutional
Neural Network using Keras API.
Architecture - CNN, transfer learning, Object
segmentation using Unet.

Securesound
- Tagged 40000+ ultrasound images for solving
object localization problem.
- Developed autonomous assisted tagging
capability in our propritear y tagging platform
using openCV.js reducing the average time taken
per batch by 30%.
- Developed a Validation platform using PHP,CSS
and javascript ,enabling professionals to validate
the tagged information with minimal effort.

Worked for an oil & gas major for maintaining and
developing their splunk cluster for SIEM and
Infrastructure Monitoring. Data onboarding from
various sources like O365 , Workday, network
devices and such onto splunk.
Splunk Certificates - Power User , Admin

-Developed python scripts to perform ETL jobs,time
range based archiving of splunk reports utilizing
Splunk's REST API.
- Configured and deployed splunk distributed
environment for creating numerous reports and on-
demand dashboards, providing insights of
performance metrices to the end users. Optimized
the performance using and automating summar y
indexing jobs.