Soham De
Bengaluru
Summary
Senior Data Scientist experienced in delivering end-to-end AI/ML solutions across Supply Chain, Revenue Management, and Process Optimization. Skilled in building scalable pipelines, leveraging GenAI/LLM, time series forecasting, and causal analysis, combining technical expertise with deep business domain knowledge. Proven track record of driving actionable insights and enabling data-driven decision making across complex enterprise operations.
Overview
5
5
years of professional experience
Work History
Advanced Data Scientist
ExxonMobil
Bengaluru
04.2024 - Current
- Led and individually contributed to the core Contract Processing team in a large-scale Supply Chain Network project, building an end-to-end AWS pipeline that applied LLMs/GenAI (prompt engineering, in-context learning, QA/QC, LLM-as-judge) for contract and invoice reconciliation, driving a projected $50M/year impact through collaboration with AWS, Accenture, and other partners.
- Revamped the Chemicals Revenue Management (CRM) tool (Delivered Value : $130M/year) by creating a scalable modeling template and developing an end-to-end forecasting solution for petrochemical markets. Applied market data, time series methods, feature engineering/selection, and ML algorithms to enhance scalability, robustness, and adoption. Improvements increased forecast accuracy, business uptake, and sales/pricing effectiveness through actionable Customer Price Recommendations and macro Price Forecasts for arbitrage insights.
Data Scientist
ExxonMobil
Bengaluru
08.2020 - 03.2024
- Causal Analysis: Applied advanced causal inference algorithms (e.g., PCMCI, FCI) for problems in energy optimization and heat exchanger fouling, combining refinery domain expertise with descriptive and prescriptive analytics. Successful field trials delivered approximately $500,000 in savings within three months across a couple of assets.
- Customer Segmentation: Developed customer clustering models using unsupervised ML to group clients into business-defined segments (Recency, Frequency, Margin) with distribution constraints, enabling tailored service-level offers, and improved business targeting.
Education
M.Tech - Chemical Engineering
Indian Institute of Technology
Mumbai07-2020
B.E - Chemical Engineering
Jadavpur University
Kolkata06-2017
Skills
- Data-driven decision making
- Machine learning and Python
- Time series analysis
- Generative AI (LLMs) & Agentic AI
- Causal inference techniques
- Customer segmentation strategies
M.Tech Thesis : ML & Bioinformatics
Built ML-based classification models to predict the efficacy of drugs against Tuberculosis (TB) strains. For a given patient's TB strain with specific mutations, the models predicted drug resistance or susceptibility, reducing reliance on time-consuming lab tests, and increasing the likelihood of effective treatment and patient survival
Accomplishments
Corporate Recognitions: Received multiple awards for successful project deliveries, leading Recruitment and providing mentorship.
Guinness World Records: Served as a trainer in the record-setting event for the largest simultaneous solar lamp lighting @ IIT Bombay, 2019.
Hobby
- Hindustani classical vocals
- Competitive quizzing
- Gardening
Languages Known
Bengali
First Language
English
Proficient (C2)
C2
Hindi
Proficient (C2)
C2
Timeline
Advanced Data Scientist
ExxonMobil
04.2024 - Current
Data Scientist
ExxonMobil
08.2020 - 03.2024
M.Tech - Chemical Engineering
Indian Institute of Technology
B.E - Chemical Engineering
Jadavpur University