Summary

Overview

Work History

Education

Skills

Languages

Timeline

Projects

ABHISHEK GUPTA

BANGALORE

Summary

Dynamic and results-driven professional with extensive experience in designing and deploying scalable solutions in cloud computing, data engineering, data science, data analytics, AI-driven platforms, and backend development. Experienced to work in health, finance, automobile, E-commerce sectors. Skilled in Python, Pandas, Pyspark, SQL, AWS and Azure servies, and cost optimization, with a proven ability to align technical implementations with business objectives. Adept at leveraging latest technologies to enhance operational efficiency and deliver measurable business value.

Overview

2

2

years of professional experience

Work History

Data Eng, Mgmt & Governance Analyst

Accenture

Bangalore

08.2022 - Current

Cognitive Insurance Platform:

CIP is a product that deals with end-to-end automation of insurance claims to minimize human intervention and maximize accuracy.
Sector: Health and Automobile.
Role: Python developer, data engineer, AWS architect, and data scientist.
Responsibility: Document ingestion from email and/or UI, Orchestration, Email-automation, Document and Email Classification.

Relevance Engine:

RE basically deals with a Unified Data Model (UDM) of an E-commerce company, and Google data (GA4, SA360, CM360, and DV360), having all the information regarding customers, products, their behavior, etc.
Sector: E-Commerce, and Google
Role: Data engineer, data scientist, data analyst, Backend developer, QA, and DevOps.
Responsibilities: The data is unstructured and raw, my responsibility was to create structured data out of UDM, data analysis on the structured data, create APIs to fetch data, check the authentication of users, test the cases, and deploy the APIs.

Generative AI Chatbot for Chief Finance Officer:

This chatbot is designed for the CFO only, to work on the Financial data of a company, the task of the chatbot is to provide financial insights to the CFO in a textual, tabular, and/or graphical format using OpenAI.
Sector: Finance
Roles: Python developer, prompt Engineer, Data Engineer, data analyst, ML Engineer, backend developer.
Responsibilities: My responsibility was to create the chatbot, fine-tune the OpenAI model, train the model using ICL; COT; and Few-shot examples, create APIs, data filtering, and UI creation for the demo.

Education

M.TECH - Mechanical Engineering

IIT PATNA

PATNA

06-2022

Skills

Cloud Services: AWS Lambda, Amazon Comprehend, Step-functions, Azure OpenAI, Lambda, S3, RDS, Event-Bridge, Azure Databricks
Data Engineering: Pandas, Pyspark, SQL, SQLAlchemy, RDS, DynamoDB, PostgresDB, MySQL, SparkSQL, Hive, Cassandra, HDFS, Scala
Data Science: Tensorflow, NLP, Deep Learning, Data Modelling, ML
Artificial Intelligence: Fine Tuning, Prompt Engineering, COT, In-context learning, GenAI, OpenAI, GPT-35, LLM, Sentence transformers, Cosine similarity

Programming Language: Python, Java, C, Scala
Business Acumen: Aligning technical solutions with business goals, cost-benefit analysis, and strategic decision-making
Project Management: Collaboration, Leadership, Problem solving

Languages

Hindi

First Language

English

Proficient (C2)

C2

Timeline

Data Eng, Mgmt & Governance Analyst

Accenture

08.2022 - Current

M.TECH - Mechanical Engineering

IIT PATNA

Projects

Cognitive Insurance Platform:

Classified submission mails on basis of priority, sentiment, business type using GenAI, Bedrock, Fine-tuning and Prompt Engineering.
Flagged the high priority mails, moved the mails to classified business type folders by using graphapi, reducing the manual effort by 90%.
Successfully led the team of 12 members, decided the AWS architecture, services to be used, and timelines for client.
Spearheaded the development of a document classification model for health sector documents using Amazon Comprehend, ensuring 50% cost reduction by transitioning to batch processing.
Automation of Amazon Comprehend model training via CLI reducing the manual effort by 99%.
Automated scaling of Inference Units based on US time zone in Amazon Comprehend Endpoints using AWS EventBridge, and lambda improving efficiency and reducing manual effort by 100%.
Implemented an image-to-PDF conversion module using the Pillow library to streamline document processing.
Automated outlook mailbox functionalities like template creation via SES, sending automated emails, folder creation, flagging and movement of mails to different folders using lambda, step-functions, SNS, SQS, used graphapi to access outlook mailbox, reducing manual effort by 90% and increasing efficiency 2x.

Relevance Engine:

Designed and deployed FastAPI-based admin APIs with JWT-based authentication, ensuring secure and scalable access to PostgresDB.
Reduced API deployment time by 50% using Docker for containerization and seamless integration into production environments.
Created and optimized database view using SQLAlchemy and Pandas, improving data processing efficiency.
Employed error handling using HTTPS module, ensuring the efficient bug fixing and future extensions.
Applied robust testing methodologies using Pytest, enhancing the reliability of APIs.

Generative AI Chatbot for Chief Finance Officer

Developed a CFO-specific chatbot using Azure OpenAI GPT-3.5, tailored to generate Python code, reducing the manual effort by 99%.
Analyzed financial data (revenue, profit margin, EBITDA) to train a model for generating Python code for finance-related analysis.
Integrated error feedback mechanisms into the chatbot, driving continuous improvement and adaptability.
Fine-tuned the AI model by providing metadata, and few-shot examples as system message on Financial terms.
Implementing ML via cosine similarity, tensorflow, scentence-transformer, and other NLP concepts for accurate query mapping, increasing the model accuracy to almost 98%.
Delivered a fully interactive demo using Streamlit, showcasing AI-powered solutions to stakeholders and securing project approval.