Lead Data scientist in Aditya Birla group. Have ample experience in predictive analysis, forecasting and Big Data technologies.
Automated process anomaly detection using LSTM Autoencoder.
Improving process yield by optimizing process parameters using NSGA-II
Working on Network optimization with constraints using integer programming
Developed Safety suit detector for high-risk areas like ARC flash (high frequency current) and Remelt areas (high temperature metal)
Developed Covid-19 Vision based use-cases: Social distancing, Mask detection, temperature measurement in a duration of 2 weeks.
This application is deployed on more than 1000 cameras across ABG
Developed android based crack detection, apparel detection (for fashion recommendation) application using tflite
Developed non-intrusive Face recognition system for 7500+ indian faces
Developed high impact(110Cr) PPE Compliance Model to detection safety helmet and safety vest violation.
Model was developed from Retinanet, Faster RCNN and SSD.
Optimized using DeepStream and TensorRT.
Automatic Number Plate Recognition module that train model independent of data for Indian License Plate
Developed non-intrusive Fire-Detection System
Developed model pipeline to predict pattern and color from image.
Model was developed with backbone of VGG19 using transfer learning.
Intrusion Detection, Unattended object detection model to automate surveillance in critical areas in the plant
Retail application such as Motion Heatmap, Interactive facial features Emotions, gaze, age gender detection.
Handled team of 5 Machine learning Engineers.
developing Computer vision products in the domain of safety surveillance and quality assurance
Automated process anomaly detection using LSTM Autoencoder
Improving process yield by optimizing process parameters using NSGA-II
Working on Network optimization with constraints using integer programming
Developed Safety suit detector for high-risk areas like ARC flash and Remelt areas
Developed Covid-19 Vision based use-cases
Developed android based crack detection, apparel detection application
Developed non-intrusive Face recognition system for 7500+ indian faces
Developed high impact PPE Compliance Model
Automatic Number Plate Recognition module
Developed non-intrusive Fire-Detection System
Developed model pipeline to predict pattern and color from image
Intrusion Detection, Unattended object detection model
Retail application such as Motion Heatmap, Interactive facial features Emotions, gaze, age gender detection
Handled team of 5 Machine learning Engineers
Unsupervised sentiment prediction model (English language only).
Used mLSTM on 84 million amazon reviews (transfer learning) to build model
Text (Blog, article, discussions) analysis for sentiment, concept, entity, summary generation using genism,word2vec and text summarizers
Managed team of 4 data Scientist, project management, project assignment and technical advisory to team, have good knowledge on handling and maintaining aws servers.
Detecting Camera Vandalism from partial full occlusion of image feed.
Facial recognition with fugitive detection in occluded condition using Dlib(CNN based implementation) with real-time accuracy of 99.9%
Realtime object detection model, implemented on High-speed car detection-based traffic model
Wrote an algorithm for vehicle counter and congestion prediction.
Algorithm worked in real-time and can be deployed on edge device.
Accident prediction using real-time weather location and road details from here-maps.
Pyspark used for implementation.
Realtime Congestion prediction from location, road surface etc. in H2o
Realtime traffic and Congestion prediction model using weather road-traffic location details.
Implementation done using H2o
Realtime Parking occupancy, noise, lighting, CO2, ambient temperature etc. Forecasting and anomaly detection
Cross domain analysis of parking and traffic, parking rating policy, lighting policy, relating noise with its root cause
Implementing Safety and Security use-cases for smart cities using data sources from Twitter.
H2o is used for model development
FPY improvement of production line by finding relating root cause of failure to actual testing procedure (Impact value – 3M USD).
PySpark is used for data engineering.
Reducing Cycle-Time by mimicking sensor using prediction algorithms.
Number of tests were reduced by generating test results based on empirical equations, developed by correlating tests.
Preventive Stock-out Inventory forecasting for large number of raw material in plants with different lead time.
Different Models is used based on lead time of each raw material
Developed data-entry error indicator.
Used Co-occurrences, clustering, outlier detection followed by anomaly detection to find error at each stage of filtering
Smart Energy balance system based on day-ahead forecasting and load management system.
This system is capable of giving flexibility load
Unsupervised sentiment prediction model
Text analysis for sentiment, concept, entity, summary generation
Managed team of 4 data Scientist
Detecting Camera Vandalism from partial full occlusion of image feed
Facial recognition with fugitive detection in occluded condition
Realtime object detection model
Algorithm for vehicle counter and congestion prediction
Accident prediction using real-time weather location and road details
Realtime Congestion prediction from location, road surface etc.
Realtime traffic and Congestion prediction model
Realtime Parking occupancy, noise, lighting, CO2, ambient temperature etc. Forecasting and anomaly detection
Cross domain analysis of parking and traffic
Implementing Safety and Security use-cases for smart cities using data sources from Twitter
FPY improvement of production line
Reducing Cycle-Time by mimicking sensor using prediction algorithms
Preventive Stock-out Inventory forecasting
Developed data-entry error indicator
Smart Energy balance system based on day-ahead forecasting and load management system
The AI-first Recruitment Operating System that unifies and automates the entire hiring lifecycle for modern enterprises worldwide.
Dr Job FZ LLC. 2026 © All Rights Reserved
