Mayank Kishore Chawla - Generative AI Data Scientist - Deloitte

Summary

Dynamic individual with hands-on experience in Generative AI and talent for navigating challenges. Brings strong problem-solving skills and proactive approach to new tasks. Known for adaptability, creativity, and results-oriented mindset. Committed to making meaningful contributions and advancing organizational goals.

Overview

2

years of professional experience

4

years of post-secondary education

Work History

DC Analyst

Deloitte

07.2023 - Current

Pioneered a secure Generative AI knowledge management system by implementing multi-vector database architecture (ChromaDB, PostgreSQL Vector, AWS OpenSearch) for specialized data types, resulting in 78% improvement in document retrieval accuracy and sub-second response times for enterprise-scale information retrieval.
Strengthened data security in Gen AI applications by implementing AWS Guardrails that prevented PII exposure, sensitive information leakage, and prompt injection attacks, achieving 100% compliance with data protection regulations while maintaining high utility of AI assistants.
Orchestrated end-to-end RAG pipelines using Langchain framework, custom AWS Claude integrations via Bedrock, and specialized document processors, resulting in a system capable of extracting insights from multi-format documents with 92% accuracy compared to human experts.
Transformed case worker operations by developing an AI copilot that translated natural language to Oracle SQL queries across 124 database tables, reducing query development time by 85% and enabling non-technical staff to independently access critical information.
Enhanced knowledge extraction capabilities by designing specialized processing pipelines combining PdfPlumber, OCR, AWS Transcribe, and large language models, enabling the system to process previously inaccessible information in legacy documents and multimedia content.
Optimized Gen AI system performance by engineering a context-aware caching layer and implementing multi-threaded retrieval architecture, reducing token consumption by 40% and cutting API costs while maintaining response quality.
Improved operational efficiency by containerizing Gen AI applications with Docker and creating standardized deployment patterns, enabling rapid scaling and consistent performance across development and production environments.
Enhanced team collaboration by providing clear communication of complex findings through visualizations and reports.
Collaborated with cross-functional teams to identify opportunities for process improvement and increased efficiency.

Education

Master Of Technology -

Indian Institute of Technology

01.2021 - Current

Bachelor of Technology - undefined

Indian Institute of Engineering Science and Technology

Higher Secondary(XII) - undefined

St. Anthony's Sr. Sec. School

Secondary(X) - undefined

Alok Sr. Sec. School

Skills

Programming Languages: Python, C, Matlab

Framework & Tools: Langchain, Streamlit, FastAPI

Generative AI Technologies: Open-source and oaid LLM models (Meta-LLAMA, Claude-Sonnet, Open-AI)

Vector Databases: ChromaDB, Pg-Vector(PostgresSQL), AWS OpenSearch (Serverless & Service)

Database Management: MySQL, MSSQL, PostgresSQL

Deployment Platforms: AWS Bedrock, AWS EC-2, Azure-ML-Studio

AI/ML Techniques: Vector embedding, NLP, Dockers, RAG, Multimodal Analysis

Soft Skills: Problem-solving, Teamwork and collaboration, Effective communication, Friendly, positive attitude, Analytical thinking

Key Projects

Covid Data Analysis | Data Exploration and Visualization | Self Project.

Data set for the Covid was collected from Kaggle community and explored to distinguish certain data in Excel.
Data Exploration and data cleaning, using several queries, were performed using Microsoft SQL server.
Various SQL VIEWS were created and Insights and Conclusions were drawn using Data Visualization in Tableau.

CIFAR- 10 Classification | Machine learning, Deep Neural Networks | Self Project

Developed a CNN architecture achieving 94% accuracy on the CIFAR-10 image classification benchmark.
Implemented data augmentation techniques and batch normalization to improve model generalization and reduce overfitting.
Optimized model training through hyperparameter tuning and learning rate scheduling, reducing training time by 35%.
Visualized network activations and feature maps to provide interpretability of classification decisions.

Accomplishments

Secured an All India Rank 20 in GATE 2021 examination.
1st position in Mowgli Challenge, Boomerang competition, INSTRUO 2018, IIEST, Shibpur.
1st position in the event, Flying Ace, ABHIYANTRIX 2020, IIEST, Shibpur.

Positions Of Responsibility

Innovation Hub Team Member, Deloitte, Present demos of cutting-edge GenAI solutions to visiting clients once monthly, showcasing Deloitte's technological capabilities and thought leadership.
Impact Day Cluster Lead, Deloitte, Volunteered as team leader during Deloitte's corporate social responsibility initiative, successfully coordinating the creation of 240 sustainable bags with creative paintings.
Head and Secretary, Aeromodelling Lab IIEST, Shibpur, Organised Aeromodelling workshop, managed the club from 2018 to 2021, responsible for all the equipment used in the lab, handled transactions, and looked up to the club's welfare.
Senior Team Lead, SAE chapter, IIEST Shibpur, Led the SAE aero design team to the national-level SAE Aero-Design competition by creating huge enthusiasm and eagerness to learn more amongst my team, responsible for fund management.

Extra-Curricular Activities

Coordinated Nvidia Deep Learning Workshop, facilitating knowledge sharing and upskilling opportunities for colleagues. Organiser and Head of RC plane-Aeromodelling workshop at IIEST, Shibpur. Successfully Piloted a few small-scale UAV aircraft post their modeling in an aeromodelling workshop.

Timeline