Arnav Somani profile photo 1 on Tal

Arnav Somani

AI & Data Science at Franklin Templeton

Bengaluru

6+ years of work experience

OPTUM (U.H.G)
Sep 20261 moPresent

Bangalore

Franklin Templeton
Sep 202511 moAug 2026

Hyderabad

Developed a Multi-Agentic RAG based solution using agentic orchestration ( LangGraph ) to retrieve relevant context, generate responses, and support model improvement through iterative fine-tuning. • Created a high-performance HR chatbot leveraging an LLM-driven RAG architecture, containerized it with Docker for deployment on Azure, achieved <0.3s response latency, and integrated the chatbot into Slack for enterprise use. • Leading a team of two AI engineers, providing technical guidance, reviewing deliverables, and ensuring high-quality execution.
Ecolab
Nov 202411 mosSep 2025

Bengaluru, Karnataka, India

• Developed and deployed a chatbot by web scraping ecolab.com (including nested URL’s), creating vector index, integrating with LLM, and serving endpoints on Databrick, then exposing to Copilot for seamless user interaction and created CI/CD pipeline for continuous updates. • Worked on creating an Agent using LangChain to facilitate multimodal data processing, enabling the identification and vectorization of diverse input sources, including video, images, PDFs and audio files, while also implementing an Agentic RAG system to enhance data retrieval and processing efficiency. • Fine-tuned the prompts to get the average accuracy score of 91% for Databricks while 90% for Pinecone while the cosine similarity score was 0.86. • Leveraged Google’s BERT to create a model for efficient label allocation of ServiceNow tickets, improving the F1-score from 73% to 87%.
Analytic Edge
Apr 20231 yr 7 mosOct 2024

• Managing end to end client engagement, from data gathering and pre-processing to model development, validation, and presentation with the end client. • Utilized OpenAi 3.5 turbo Large Language Model (LLM) to create a chatbot for a fortune 1000 electronic company using RAG ( Retrieval Augmented Generation ) and also fine-tuned the LLM using Low Rank Adaptation fine-tuning. • Worked with Fortune 500 clients, to help build predictive models and optimize marketing strategies using statistical technique like Marketing Mix Model and machine learning method like regression. • Build interactive propensity model in python using XGBoost Classifier (ML Technique) to predict the likelihood of the users which may be most responsive to SMS marketing resulting in an increase of conversion ratio to 12% from 8%. • Using python predicted the Lifetime Value (LTV) of a subscriber of an OTT platform using survival techniques like curve fitting method, Cox Proportional Hazard (CPH) regression.
Analytic Edge
Apr 20221 yrMar 2023

Analytic Edge
Jun 202110 mosMar 2022

ARCHER Systems, LLC
Sep 202010 mosJun 2021

Vadodara, Gujarat, India

• Responsible for the extraction, manipulation, and analysis of client data for internal and external stakeholders, using SQL, Excel and internal tools. •Implementation of the SQL queries for the identification of data discrepancies. Leading a team of 3 Analyst to execute and prioritize workflow, managing and processing the data while increasing the efficiency by 65%. • Working on a script for the automation of data retrieval from database to a modified excel sheet ready for reporting via SQL and Python
Clean Energy Associates (CEA)
Apr 20205 mosAug 2020

Shanghai, China

• Support CEA's Sales and Marketing Team by providing on-demand MI analysis and data visualizations using python and PowerBi. • Collect, process, cleanse and organize market intelligence (MI) data from solar and storage sectors. • Support development and implementation of MI data management and visualization via software platforms, such as Power BI, SharePoint and O365 applications. • Process and organize CEA's archived data using EDA and data cleaning techniques in python. • Create client facing content with customized visualizations and supporting analysis using python. • Support maintenance and further development of CEA's MI subscription products which include real time benchmarking statistics and periodic report publications.

Arnav’s AI Stack

antigravity
antigravity
claude code
claude code
codex
codex
deepseek
deepseek
elevenlabs
elevenlabs
grok
grok

Education

New York University

Master's degree, Management of Technology

New York University

20152 yrs2017
International Institute of Information Technology Bangalore

PG Diploma, Machine Learning and Artificial Intelligence

International Institute of Information Technology Bangalore

20201 yr2021

Join Arnav and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Arnav Somani on Tal

Headline

AI & Data Science at Franklin Templeton

Location

Bengaluru

AI stack

antigravity

claude code

codex

deepseek

elevenlabs

grok

Work experience

Senior AI/ML Engineer at OPTUM (U.H.G), Sep 2026 - Present · 1 mo, Bangalore

Senior AI/ML Engineer at Franklin Templeton, Sep 2025 - Aug 2026 · 11 mo, Hyderabad, Developed a Multi-Agentic RAG based solution using agentic orchestration ( LangGraph ) to retrieve relevant context, generate responses, and support model improvement through iterative fine-tuning. • Created a high-performance HR chatbot leveraging an LLM-driven RAG architecture, containerized it with Docker for deployment on Azure, achieved <0.3s response latency, and integrated the chatbot into Slack for enterprise use. • Leading a team of two AI engineers, providing technical guidance, reviewing deliverables, and ensuring high-quality execution.

Senior Data Scientist at Ecolab, Nov 2024 - Sep 2025 · 11 mos, Bengaluru, Karnataka, India, • Developed and deployed a chatbot by web scraping ecolab.com (including nested URL’s), creating vector index, integrating with LLM, and serving endpoints on Databrick, then exposing to Copilot for seamless user interaction and created CI/CD pipeline for continuous updates. • Worked on creating an Agent using LangChain to facilitate multimodal data processing, enabling the identification and vectorization of diverse input sources, including video, images, PDFs and audio files, while also implementing an Agentic RAG system to enhance data retrieval and processing efficiency. • Fine-tuned the prompts to get the average accuracy score of 91% for Databricks while 90% for Pinecone while the cosine similarity score was 0.86. • Leveraged Google’s BERT to create a model for efficient label allocation of ServiceNow tickets, improving the F1-score from 73% to 87%.

Associate Manager (Data Scientist) at Analytic Edge, Apr 2023 - Oct 2024 · 1 yr 7 mos, • Managing end to end client engagement, from data gathering and pre-processing to model development, validation, and presentation with the end client. • Utilized OpenAi 3.5 turbo Large Language Model (LLM) to create a chatbot for a fortune 1000 electronic company using RAG ( Retrieval Augmented Generation ) and also fine-tuned the LLM using Low Rank Adaptation fine-tuning. • Worked with Fortune 500 clients, to help build predictive models and optimize marketing strategies using statistical technique like Marketing Mix Model and machine learning method like regression. • Build interactive propensity model in python using XGBoost Classifier (ML Technique) to predict the likelihood of the users which may be most responsive to SMS marketing resulting in an increase of conversion ratio to 12% from 8%. • Using python predicted the Lifetime Value (LTV) of a subscriber of an OTT platform using survival techniques like curve fitting method, Cox Proportional Hazard (CPH) regression.

Data Scientist at Analytic Edge, Apr 2022 - Mar 2023 · 1 yr

Analytics Consultant at Analytic Edge, Jun 2021 - Mar 2022 · 10 mos

Data Analyst at ARCHER Systems, LLC, Sep 2020 - Jun 2021 · 10 mos, Vadodara, Gujarat, India, • Responsible for the extraction, manipulation, and analysis of client data for internal and external stakeholders, using SQL, Excel and internal tools. •Implementation of the SQL queries for the identification of data discrepancies. Leading a team of 3 Analyst to execute and prioritize workflow, managing and processing the data while increasing the efficiency by 65%. • Working on a script for the automation of data retrieval from database to a modified excel sheet ready for reporting via SQL and Python

Analyst at Clean Energy Associates (CEA), Apr 2020 - Aug 2020 · 5 mos, Shanghai, China, • Support CEA's Sales and Marketing Team by providing on-demand MI analysis and data visualizations using python and PowerBi. • Collect, process, cleanse and organize market intelligence (MI) data from solar and storage sectors. • Support development and implementation of MI data management and visualization via software platforms, such as Power BI, SharePoint and O365 applications. • Process and organize CEA's archived data using EDA and data cleaning techniques in python. • Create client facing content with customized visualizations and supporting analysis using python. • Support maintenance and further development of CEA's MI subscription products which include real time benchmarking statistics and periodic report publications.

Education

Master's degree, Management of Technology, New York University

PG Diploma, Machine Learning and Artificial Intelligence, International Institute of Information Technology Bangalore