Open to opportunities

Arpan Somani

@arpansomani

Message

Applied scientist and ML engineer specializing in GenAI, LLMs, and image recognition for defense and enterprise.

India

Message

What I'm looking for

I seek roles applying ML/LLM/GenAI to impactful, mission-driven problems where I can lead deployment of secure production systems and advance model reliability and scalability.

I am an Applied Scientist with 8+ years of experience building and deploying AI, ML, LLM and deep-learning solutions in industry and government settings.

At Amazon I led multilingual LLM and NLP work for Alexa production, fine-tuned models for Indian languages, and shipped models that improved customer metrics. I designed and deployed internal LLM chatbots and automated document summarization pipelines.

In my current role I lead execution of GenAI and image-recognition projects for the Ministry of Defense, built RAG systems with secure vector stores, and delivered source-grounded responses from classified documents.

I hold an M.Tech in Computer Science from IIT Bombay, have published and patented work in NLP and entity linking, and seek to apply practical research and robust engineering to mission-critical AI systems.

Experience

Work history, roles, and key accomplishments

Applied Scientist II

Amazon

Nov 2020 - Apr 2024 (3 years 5 months)

Led development and deployment of GenAI and image-recognition solutions for the Ministry of Defence, built an internal LLM-powered chatbot and RAG systems, and implemented document summarization pipelines that accelerated decision-making.

Large Language Models PyTorch NLP Summarization Vector Databases Image Recognition

Research Engineer

American Express

Aug 2018 - Oct 2020 (2 years 2 months)

Developed ML pipelines for record linkage and entity-matching for credit fraud detection, saving approximately $6M annually and built a hybrid entity linking system combining BERT-based semantics and syntactic features.

BERT WMD XGBoost PyTorch Solr Elasticsearch Fraud Detection

Application Developer

Oracle

Jul 2017 - Aug 2018 (1 year 1 month)

Implemented semantic search enhancements for a B2B procurement app, developed APIs, optimized code and SQL for Oracle's cloud procurement application.

Semantic Search APIs SQL C++Shell Scripting Information Retrieval

Education

Degrees, certifications, and relevant coursework

Indian Institute of Technology Bombay

Master of Technology, Computer Science

2015 - 2017

Grade: CPI 9.22

Activities and societies: Graduate Teaching Assistant for Web Search & Mining and Data Structures & Algorithms; Interview Coordinator at Placement Cell.

Completed a Master of Technology in Computer Science with a CPI of 9.22, involving graduate coursework and research projects in machine learning and NLP.