Summary
Overview
Work History
Education
Skills
Some of the Selected US Patents (10+)
Some of the renowned Awards
Some of the recent Publications (30+)
Timeline
Generic
Amit Agarwal

Amit Agarwal

Bangalore

Summary

An accomplished AI leader with over 10+ years of experience in conversational and generative AI, specializing in NLP, ASR, machine learning, and advanced applications such as recommendation systems and time series forecasting. Holds a PhD degree in a related field, with a robust background in strategic and financial management of AI initiatives. Proven ability in architecture design, solution development, team leadership, and innovation, with a focus on aligning AI technologies with business objectives. Skilled in stakeholder management, ethical AI practices, and fostering a culture of excellence.

Overview

10
10
years of professional experience

Work History

Senior Research Scientist

Wells Fargo India
03.2021 - Current
  • Credit Memo Generation and Evaluation (Commercial Banking LOB): Led a project leveraging LLMs such as the Gemini model to summarize credit memos based on financial workbooks and auditor notes. The project addressed an annual requirement of ~5,000 credit memos derived from ~15,000 documents, each previously requiring 26-40 hours for summarization. The solution is projected to save ~$23MM annually, improving client and employee experiences by:Reducing annual review cycle times by 80%.
    Reducing renewal and extension cycle times by 40%.
  • LLM-Powered Summarization for Customer Calls (CSBB LOB): Directed a project utilizing LLMs to summarize customer calls, addressing context window challenges. This initiative demonstrated enhanced operational efficiency, with an estimated impact of $3.45 million in cost savings.
  • Agentic AI Interview Screening System (Pilot) – Wells Fargo, CSBB:Led the development of an Agentic AI–driven candidate screening system to automate the first-round interview process for CSBB hiring. The system combined LLM-based conversation agents, structured evaluation rubrics, and scoring workflows to assess communication ability, role fit, and domain understanding. Impact: Reduced average recruiter screening time from approximately 25 minutes per candidate to approximately 3 minutes, resulting in an approximately 88% reduction in effort. Cost saving $ 176K.
  • Audit Policy ChatBot Development: Directed an Information Retrieval project using Flan-T5, BART, Llama-2, and BERT models for question generation and answer retrieval. Developed a ChatBot integrated with audit documentation, demonstrating proficiency in generative AI to streamline audit policy reviews. Project impact: $1.5 million.
  • Complaint Resolution Classification Model (POC): Development: Created achallenger model using BERT for topic classification, which identifies potentially easy resolution cases and routes them to specific resolution teams. This model achieved an 8.2% increase in accuracy compared to the baseline model developed by the CEDA team.
  • Regulatory Document Analysis and Data Quality Assurance :Project Scope : Developed an Information Retrieval solution to process regulatory documents (e.g., Mortgage loan documents) stored as PDFs or images. Implemented data extraction techniques on text outputs from the KOFAX tool to identify data quality gaps by comparing against SOR values.Current Status: Fully deployed and operational within production environments.
  • Automated Information Extraction for Loan Processing(POC):Challenge:Addressed the time-consuming and error-prone process of manually reviewing and matching loan document details, regulated under RESPA/TILA—posing high risks for Wells Fargo.
    Solution: I have developed an automated information extraction solution to validate customer details across various document types, enhancing efficiency and accuracy in manual validations. This output supports the NNCE team's efforts to improve process accuracy.
  • Session-Based Recommendation Systems: Designed and implemented a session-based recommendation system for the MovieLens dataset, utilizing state-of-the-art architectures such as HRNN and TR4REC to deliver personalized movie recommendations.
  • Sales Forecasting for Inventory Optimization: Developed a sales forecasting model for the M5 Walmart dataset, enabling accurate future sales predictions and optimized inventory management.

Senior Data Scientist

Wells Fargo India
01.2020 - 03.2021
  • As a Senior Data Scientist, I spearhead collaboration with the Innovation team, focusing on identifying key problem areas and crafting groundbreaking solutions. My role encompasses the development of proofs of concept (POC), closely followed by coordination with the legal team to ensure the uniqueness and patentability of these innovations. My contributions have led to the successful filing of 10 US patents and the publication of over 20 research papers, underlining my commitment to advancing the field of data science through strategic innovation and legal safeguarding of intellectual property.

Research Scholar

IIT Roorkee Data Mining Lab
01.2016 - 12.2019

Student Trainee Internship

Samsung Research Institute Bangalore
01.2018 - 05.2018
  • I have worked on NER based model on chat data

Education

PhD - CSE Department (Best Thesis Award)

Indian Institute of Technology Roorkee
Roorkee, IN-UK
2020

Master of Technology (M.Tech.) - CSE Department

Graphic Era University
Dehradun, IN-UK
2015

B.Tech - CSE Department

College of Engineering Roorkee
Roorkee, India
2012

Skills

  • NLP, Generative AI (summarisation, information retrieval from PDF), RAG, agentic AI, Langchain, LangGraph
  • Agentic AI, MCP
  • Gemini Pro, Chat GPT 35, Llama, Whisper, Mistral, Roberta, etc
  • LLM optimisation by utilising Parameter Efficient Fine Tuning (PEFT) - LoRa, QLoRa, Dora
  • Knowledge distillation, causal inference
  • Speech-to-Text (GenAI & Medical)
  • FastAPI, Streamlit, Docker, GitHub, and GCP
  • Time-series forecasting model (Walmart sales prediction)
  • Recommendation Systems
  • Good experience in stakeholder interaction

Some of the Selected US Patents (10+)

  • Complaint Prioritization using Deep Learning Model, 1234-235US01, 06/03/2021
  • Predicting Customer Interaction Using Deep Learning Model, 1234-236US01, 06/25/2022
  • Emotion Analysis using Deep Learning Model, 1234243US01, 11/23/2022
  • Fraud Detection Using Emotion Based Deep Learning Model, 1234-244US01, 08/09/2022
  • Smart Call Routing Using Deep Learning Model, 1234254US01, 01/13/2023

Some of the renowned Awards

  • Best Innovator Award (2023): Awarded by Wells Fargo India for leading innovative projects to successful completion, exemplifying cutting-edge solutions and strategic leadership in AI.
  • AI Leadership Award (2023): Recognized for outstanding leadership in AI, driving forward-thinking strategies and the development of high-impact AI solutions that align with business goals
  • Received Innovation of the year award in 2022 at Wells Fargo for most US Patent filed.
  • Invited AI Expert Speaker: Featured at prestigious academic institutions such as IIT-R, NIT-M, and Thapar University, sharing insights on the latest AI trends and research findings.
  • DAAD Fellowship for ASIRF-2019
  • Microsoft Travel Grant to present a paper in ECML/PKDD'19 Wurzburg, Germany
  • Google Scholarship for attending LxMLS-2017 (Portugal)
  • ACM SIGCHI Scholarship for Summer School-2019 Barcelona (Spain)

Some of the recent Publications (30+)

  • "Enhancing Medical Transcription and Correction using transformer-based ASR and Fine tuned LLM" accepted in The International Conference on Knowledge Discovery and Data Mining A* conference (KDD 2024), Spain
  • "AgriLLM: Harnessing Large Language Models for Farmer Queries" accepted in The International Conference on Knowledge Discovery and Data Mining A* conference (KDD 2024), Spain
  • "Snowy Scenes,Clear Detections: A Robust Model for Traffic Light Detection in Adverse Weather Conditions" accepted in The International Conference on Knowledge Discovery and Data Mining A* conference (KDD 2024), Spain
  • "IITRoorkee@SMM4H 2024 Cross-Platform Age Detection in Twitter and Reddit Using Transformer-Based Model" accepted inThe 9th Social Media Mining for Health Research and Applications (#SMM4H) Co-located with (ACL 2024), A* conference.
  • "Unveiling Themes in Judicial Proceedings: A Cross-Country Study Using Topic Modeling on Legal Documents from India and the UK" accepted in 14th International Conference on Formal Ontology in Information Systems (FOIS 2024) A* conference
  • Identifying Leadership Characteristics from Social Media Data During Natural Hazards Using Personality Traits, Scientific Reports - Nature, SCI Q1, 4.9

Timeline

Senior Research Scientist

Wells Fargo India
03.2021 - Current

Senior Data Scientist

Wells Fargo India
01.2020 - 03.2021

Student Trainee Internship

Samsung Research Institute Bangalore
01.2018 - 05.2018

Research Scholar

IIT Roorkee Data Mining Lab
01.2016 - 12.2019

PhD - CSE Department (Best Thesis Award)

Indian Institute of Technology Roorkee

Master of Technology (M.Tech.) - CSE Department

Graphic Era University

B.Tech - CSE Department

College of Engineering Roorkee

Similar Profiles

CREATE PROFILE
Amit Agarwal