Summary
Overview
Work History
Education
Skills
Publications
Accomplishments
Timeline
Generic
NVS Abhilash

NVS Abhilash

Machine Learning Engineer II
Bangalore

Summary

Experienced Machine Learning Engineer with 7 years of strong background in training, optimizing, and deploying deep learning and computer vision models at scale. Skilled in utilizing cloud infrastructure to enhance efficiency and performance, contributing to successful project outcomes.

Overview

7
7
years of professional experience
4
4
years of post-secondary education

Work History

Machine Learning Engineer II

Amazon
09.2022 - Current
  • Designed and implemented an optimized cluster centroid proposal job in Python using joblib and pandas. This initiative replaced an existing process, resulting in a runtime reduction of 7 days and slashing cloud costs by 100x.
  • Developed and deployed a high-accuracy Image Verification model using PyTorch, integrating it into the production inference pipeline.
  • Optimized model architectures and training strategies, while streamlining the team's training and orchestration framework, reducing the model development lifecycle from 3 days to 1 day.
  • Implemented a Deep Learning evaluation pipeline in Python, leveraging PyTorch, to enhance the evaluation process of a image verification model. This pipeline involves data retrieval, inference, and metric generation, streamlining the overall process.
  • Independently developed a faster offline clustering solution using HNSW, replacing the previous inefficient system. This resulted in saving 5 days of runtime and reducing costs by 300x.
  • Designed and implemented an 'instant alert' notification and validation service to automatically evaluate edge cases of our ML system by running heavier Deep Learning models within 5 minutes of the event using AWS Lambda and Docker.
  • Managed and enhanced a Root Cause Automation job utilizing deep learning techniques to automatically annotate failure instances within our ML system, improving efficiency and accuracy, resulting in saving 10-15 hours of human effort per week.

Data Scientist

Genpact
07.2018 - 09.2022
  • Implemented a Computer Vision based AI pipeline involving image classification and semantic segmentation models to reduce time involved vehicle insurance claims process from 1-2 weeks to 2-3 hours.
  • Implemented end-to-end optimized machine learning pipeline for production deployment using Python OOPs, Flask, and Docker.
  • Trained, deployed, and monitored NLP models for textual classification and topic modeling use cases for multiple clients.
  • Automated machine learning pipelines leveraging MLOps in Azure Machine Learning to improve the experiment iteration time by 20% of a project and save development costs by 40%.
  • Mentored 5 Data Science Interns for projects involving Computer Vision, AutoML, and Active Learning (weak supervision-based approach).
  • Filed 2 patents and co-authored 3 research papers for approaches involved in building Computer Vision modules for damage detection.

Education

Bachelor of Technology - Computer Science and Engineering

NIIT University
Neemrana, Rajasthan
07.2014 - 10.2018

Skills

Python, C/C, SQL

PyTorch, Triton, CUDA

Git, Linux, Docker

MLOps, CI/CD, PySpark

AWS Serverless Stack (S3, Lambda, Batch)

AWS Sagemaker

undefined

Publications

Real time Chest X-ray Pathology detection and localization framework with Convolutional Neural Networks and Ensembling
2022 International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME), Maldives, Maldives, 2022
IEEE Publication: https://ieeexplore.ieee.org/document/9988570

Duplicate Instance Identification In Multiview And Multiscale Systems
2020 IEEE Sixth International Conference on Multimedia Big Data (BigMM), New Delhi, India, 2020
IEEE Publication: https://ieeexplore.ieee.org/document/9232529

Accurate Damage Dimension Estimation in AI Driven Vehicle Inspection System
7th National Conference, NCVPRIPG 2019, Hubballi, India, December 22–24, 2019
Springer Publication: https://link.springer.com/chapter/10.1007/978-981-15-8697-2_14

Online Partitioning of Large Graphs for Improving Scalability in Recommender Systems
2017 International Conference on Computational Intelligence: Theories, Applications and Future Directions, December 6-8, 2017
Springer Publication: https://link.springer.com/chapter/10.1007/978-981-13-1135-2_10
Github Source code: https://github.com/nvs-abhilash/OSNTrust

Accomplishments

Dimension estimation using duplicate instance identification in a multiview and multiscale system

Published on 2024/01 in US PTO - ACTIVE

https://patents.google.com/patent/US20230063002A1


Artificial Intelligence Based Determination Of Damage To Physical Structures Via Video

Published on 2024/07 in US PTO - ACTIVE

https://patents.google.com/patent/US20220129860A1


System and method for artificial intelligence based determination of damage to physical structures

Published on 2022/08 in US PTO - ACTIVE

https://patents.google.com/patent/US11875496B2

Timeline

Machine Learning Engineer II

Amazon
09.2022 - Current

Data Scientist

Genpact
07.2018 - 09.2022

Bachelor of Technology - Computer Science and Engineering

NIIT University
07.2014 - 10.2018

Similar Profiles

CREATE PROFILE
NVS AbhilashMachine Learning Engineer II