Somil Gupta Email & Phone Number

Mumbai, Maharashtra, India

Tokyo, Japan

Led the development of Helpdesk platform to automate ticket resolution, ensuring context-aware responses by integrating an advanced GenAI & RAG pipelines.Automated Redmine ticket polling and asynchronous processing, significantly reducing manual intervention and response time.Tech Stack: Next.js, FastAPI, Azure Cloud, Kafka, VectorDB, Knowledge Graph, Multi-hop RAG, ReAct.

Tokyo, Japan

Built AI-driven Multi Functional Technology (MFT) Analysis pipelineEngineered and deployed an MFT analysis pipeline to support R&D decisions by automating data retrieval and processing.Fine-tuned transformer models and set up an automated data pipeline, delivering insights through a fully integrated cloud-based solution.Tech Stack: Docker, AWS Batch, ECR/ECS, CloudFormation, AWS SageMaker, T5 transformer, AWS S3, Lambda, Fargate, QuickSight.

Bengaluru, Karnataka, India

Developed an ML-based expense categorization system for a smart financial management app, improving accuracy and reducing manual efforts.Built a merchant identification pipeline using advanced NLP techniques, streamlining transaction classification.Tech Stack: BERT, Spacy NER, regex, KMeans clustering, TF-IDF vectors, PyTorch

Validated and enhanced the Operational Risk Stress Testing Model by independently testing and improving its accuracy.Addressed documentation gaps and automated key processes to ensure robust model validation and compliance.Tech Stack: R, statistical testing, model validation techniques.

Bangalore

Created an AI-driven Early Warning System to identify underperforming sales reps using internal sales data and metrics.Implemented a hybrid model for time series forecasting, ensuring accurate predictions across various scenarios.Tech Stack: Hybrid stacked models, time series forecasting, statistical and ML methods.

Mumbai Area, India

Developed a ranking system for M&A rumors using machine learning, enhancing the accuracy of rumor classification.Analyzed and improved NER models to retrieve and classify entities from financial news sources.Tech Stack: Random Forest, XGBoost, BiLSTM, Residual Networks, CRF, GloVe, ELMo embeddings.

Bengaluru Area, India

Analyze the Market Impact of Public Release of Voting Proposals in Equity Mutual FundsConducted sentiment analysis on 150k individual votes for proposals in AGMs of 1500 Indian firms over three years, categorizing the sentiment and its implications.Applied predictive models in Python to assess the market impact following the public release of voting outcomes, and developed a web scraper with Selenium and VBA for data retrieval and preprocessing.Tech Stack: Python, Sentiment Analysis, Predictive Modeling, Selenium, Chrome Driver, VBA.

Dual Degree, Financial Engineering

B.Tech, Chemical Engineering

High School And Intermediate, Science, A