Amit Pandey
Data Scientist
Hyderabad
Summary
Data Scientist with over 5 years of experience leveraging machine learning, statistical modeling, and Generative AI to solve complex business problems. Holds a Master's in Data Science and a Bachelor's in Mathematics, providing a strong theoretical and practical foundation. Proven expertise in the end-to-end project lifecycle, including client delivery and project management. Seeking to apply advanced analytics skills in a Data Scientist,Ml & GenAI engineer, Advanced Analytics Consultant, or Marketing Science role.
Overview
5
5
years of professional experience
3
3
Certifications
Work History
Data Scientist(AM)
EXL
Hyderabad
04.2024 - Current
- Tools & Skills: Databricks, PySpark, Phi3 & LLM finetuning, Machine learning, forecast, A/B testing, CI/CD, GitHub action, MLOps, Marketing Mix modeling, Marketing Science, GenAI, Agentic AI, Copilot studio,Hybrid search
- Client- One Main Financial(OMF), VISA, Dow Jones
- Built an Agentic AI chatbot using copilot studio to extract consumer insight from LAM market research reports, reducing manual analysis effort, >=95% of generative answers include 1 + citations to approved sources, >60-70% reduction in time to first draft, 3x to 5x increase in reuse of existing research content across campaigns
- Built and deployed Media Mix Models (Lightweight MMM, PyMC, Ridge, RF, SHAP) on Databricks with MLflow and CI/CD (GitHub Actions), enabling automated retraining, logging, and monitoring; delivered +15% forecast accuracy and 5-10% marketing spend efficiency through actionable budget allocation and incrementality insights.
- Developed a sentiment classification prototype for a banking client using fine-tuned open-weight LLMs (Meta LLaMA-3B, Microsoft Phi-3.3B) on financial news/document headlines
- Applied Parameter-Efficient Fine-Tuning (LoRA/SFT) on GPU with prompt-based multi-class training, boosting Phi-3 accuracy from 75.1% to 87.4% and LLaMA-3 from 57.6% to 85.8%
- Built a hybrid business name matching system leveraging syntactic (character-level string similarity via BM25) and semantic (Sentence-BERT + FAISS) techniques to resolve noisy and ambiguous entity names.
Asst. Manager- Advance Analytics/Marketing Science
IPSOS
Bangalore
06.2023 - 04.2024
- Client- Swarovski, Philip Morris International & O'Reilly
- Marketing Mix Modeling, Multi-Touch attribution, Unified attribution modelling, Retail & CPG, Client Solution/Management Group, Project management, Team lead, analytics lead, Marketing Science, Data Science, forecasting, Optimization, Retail &CPG, automobile
- Refined the Brand-to-Performance media mix (66:34) across US and EU markets, successfully safeguarding a €191M revenue stream while identifying high-growth channels to offset a 35% investment reduction
- Identified a 1.6x higher ROI synergy between media and non-media activities, utilizing MMM insights to reallocate budget and drive a 10% YoY revenue increase in the retail channel
- Leveraged Marketing Mix Modeling (MMM) to quantify short-term ($379MM) vs. long-term ($300M-$450MM) revenue impact, validating total marketing contribution of over $680MM
- Utilized advanced optimization modeling to achieve a $21.4MM net sales lift by strategically reallocating spend to high-efficiency digital tactics (Search/Display), and successfully preventing a potential $15MM revenue erosion from proposed budget reductions
Data Scientist-Product(NLP/Gen AI)
1digitalstack.ai
Gurgaon
03.2023 - 06.2023
- E-Commerce Science, Digital shelf management, NLP, Topic Modeling, Custom Named Entity Recognition, Multi-class text classification, sentiment classification
Senior Executive-Advance Analytics/Marketing Analytics
GroupM
Gurgaon
02.2023 - 03.2023
- E-Commerce Science, Digital shelf management, NLP, Topic Modeling, Custom Named Entity Recognition, Multi-class text classification, sentiment classificationClient- Facebook, British Petroleum, Mazda, Unilever, TCS, ASU & Blue Buffalo
- Methodology- Performance modeling, an agile & predictive version of Marketing Mix Modeling (MMM), Multivariate Forecasting, Multi label Text classification, Topic modeling.
- Collaborating with data scientists across continents and delivering business value through ML projects for Fortune 500 clients.
- Reporting to the Global Director of Advanced Analytics of the MindShare agency of GroupM.
- Quantified media effectiveness across digital investments to drive optimization and forecasting. Optimized media to achieve given business targets, Predicted upper & lower funnel KPIs given media budgets.
- Run different budget scenarios based on incremental or budget cuts to determine which channels contribute to KPI performance based on ML based model.
- The forecasted KPIs have supported a full year of media planning and reduced Cost Per Purchase by 50% YOY on the application.
- The model output helps the strategy team achieve growth in purchase rates by 50% YOY.
Associate Data Scientist
Great Learning
Gurgaon
12.2020 - 02.2022
- NLP, Text classification, statistical modelling, Acad ops, forecasting, NER, EMachine learning project life cycle
Education
M.Sc. - Data Science
MUIT
Noida01.2020
B. Sc. - Mathematics
Kallinga University
01.2016
Skills
- Agentic AI, RAG & LLM fine tuning
- Machine learning, Statistical modeling & Forecasting
- AI/MLOPS
- Experiment Design(A/B testing)
- Marketing Mix modeling & A/B testing
- Databricks, Snowflake
- Python , SQL & Pyspark
- Project Management
Websites
Certification
Marketing Analytics - Darden School of business
Timeline
Data Scientist(AM)
EXL
04.2024 - Current
Asst. Manager- Advance Analytics/Marketing Science
IPSOS
06.2023 - 04.2024
Data Scientist-Product(NLP/Gen AI)
1digitalstack.ai
03.2023 - 06.2023
Senior Executive-Advance Analytics/Marketing Analytics
GroupM
02.2023 - 03.2023
Associate Data Scientist
Great Learning
12.2020 - 02.2022
M.Sc. - Data Science
MUIT
B. Sc. - Mathematics
Kallinga University