Prabhat Kumar

Control Management Machine Learning (CMML):

• Built Question Answer and classification model using BERT, Roberta and other transformer architecture for risk, process, issue, business, and legal descriptions.
• Worked on deep hierarchical models for legal entity recommendation for internal processes, risks and controls.
• Worked on clear language model for complex and simple sentence detections. Also built jargon detection NER model as part of this use-case.
• Involved in building regulatory topics and obligations to controls mapping models.
• Built advanced hybrid retrieval techniques using Sparse retrievers, embedding based retrievers and Knowledge graph based retrievers.
• Experienced in training adapter models for specialized tasks using llama index ,open-ai and sentence transformer.
• Created solution for PII detection using transformers and Microsoft Presidio.
• Worked on fine-tuning of dense passage retriever (DPR) and reader model using haystack framework.
• Worked on QA, Classification, NER and Seq2Seq modelling and its training.
• Experienced in customizing transformer architectures, creating its dataloaders and prediction pipeline using pytorch lightning, pytorch and transformers package.
• Worked with technology to design machine leaning inference architecture based on AWS.

Cognitive Banking - Project UNO

• Partnered in development of award winning Hyper-personalized AI recommendation system for 200+ retail banking products for DBS Singapore retail customers (Product Acquisition , Usage , Channel, Content and Time recommendation models) with business impact of 25 million SGD.
• Responsible for feature engineering, selection, training and optimization of 200+ MMLSpark /SnapseML based models on spark cluster.
• Created pyspark-based python package for doing feature engineering and selection, parallel hyper tuning, parallel training, evaluation, and inference.
• Involved in auto tagging and extraction of information using Vision transformer, Faster R-CNN , Masked R-CNN and AWS Rekognition from images used for DBS marketing campaigns.
• Played key role in building models for review categorization and sentiment classification of DBS apps reviews on Appstore and playstore using traditional NLP and Transformers.
• Involved in solving optimization problem using constraint programming and discrete optimization.
• Built risk score model for cheques and optimization model for allocation of cheques (Operation research) with a business impact of 500K SGD.
• Utilized NLP techniques like zero-shot learning in content model and feature discovery phase.
• Experienced in modelling with xgboost, catboost, mmlspark lightgbm (snapseml), lightgbm (non- spark), and neural net architectures like tabnet, tab transformer, node, etc.
• Created experiments for doing unsupervised learning and then reusing layers in supervised learning with labeled data.
• Experienced in using hyper-parameter tuning libraries like optuna, hyperopt in both non-spark and spark environments for traditional ML and deep learning use cases
• Involved in people management tasks like mentoring and guiding resources, taking feedback, and ensuring that their career is on right path.
• Also involved in various POCs related to NLP, Classification, and Speech Analytics and also participated in various hackathons on Analytics Vidhya and HackerEarth ranging from topics like classification, regression, image classification, and NLP

• Extensively worked on Performance Engineering and ETL projects.
• Experienced in C Programming and deploying them to HP Performance center for performing load testing on servers
• Use statistical methods for extrapolating performance metrics of Tech test server to production servers using R.
• Experienced in developing PL/SQL code for generating data for various reports.

• Worked in US Healthcare domain
• Analysis of error on production and perform replication in QA environment
• Involved in analysis of new requirements and provide user requirement documents to developers for coding
• Creating Business Acceptance test-case for final acceptance
• Involved in reviewing code and suggesting any changes if required.

• Extensively worked on IBM Power System (IBM i), PL/SQL, SQL, and campaign management in telecom domain
• Implemented KPIs using SQLs and PLSQL code
• Involved in creation of segment, sub segment and campaigns using KPIs.
• Tracking performance of campaigns and it subsequent optimization.