I am a Senior Retail Professional and Category Head with over 24 years of industry experience, currently specializing in Ecommerce, Modern Trade, Merchandising, Agentic AI and Data Science. I bridge the gap between high-volume retail operations and autonomous AI engineering, building systems that translate raw data into instant, prescriptive insights.

🧠 Current Focus: I am alumnus of the Indian Institute of Foreign Trade, New Delhi and possess Professional Certificate Programme in Agentic AI and Data Science at IIT Madras. Also holding AI certifications from Google, Microsoft, IBM, and AWS. I am a Law Graduate from the University of Delhi and multiple certifications in Physics, AI, Data Science and Terchnology.

🚀 Featured Retail AI Architecture I have built a complete, end-to-end retail analytics suite, transitioning from traditional predictive modeling to autonomous LangChain agents:

• Demand Sensing & Probabilistic Forecasting: A Temporal Fusion Transformer (TFT) architecture utilizing PyTorch Forecasting and Quantile Loss (P10/P50/P90) for dynamic retail safety stock optimization.
• Automated Catalog Classifier: A fine-tuned DistilBERT NLP model hosted via Hugging Face that autonomously maps raw vendor text into internal retail categories.
• Two-Tower Neural Recommendation Engine A custom PyTorch architecture utilizing in-batch negative sampling and sub-millisecond FAISS vector search for real-time candidate retrieval.
• Retail Agentic AI: Text-to-SQL Assistant : A localized LangChain and Gemini 2.5 Flash agent that bypasses manual dashboarding by autonomously translating plain English operational queries into executed SQL against a simulated MySQL environment.
• Yield Optimization Engine : A dynamic mathematical API calculating price elasticity to minimize spoilage and maximize gross margin for highly perishable inventory.
• Customer Segmentation API : An unsupervised machine learning backend utilizing RFM clustering to group shoppers into actionable behavioral segments.
• Demand Forecasting API : A production-ready FastAPI service leveraging Random Forest Regression for inventory optimization.
• Review Sentiment & Triage : An NLP microservice that classifies customer feedback and automatically routes operational incident tickets.

🌱 Additional Projects

• SBI Financial Database : A comprehensive SQL architecture featuring optimized views and an integrated Power BI dashboard for enterprise reporting.
• Spendly Expense Tracker : A lightweight, full-stack personal finance Flask web application built to log daily expenses and visualize spending patterns.
• Ames Housing ML Pipeline : An end-to-end machine learning pipeline predicting real estate prices using demographic and geographical feature engineering.

🛠️ Technical Stack

• Languages & Frameworks: Python, SQL, FastAPI, LangChain, LangGraph, PyTorch
• AI & Machine Learning: Google Gemini, Hugging Face, Transformers, PyTorch Forecasting, Random Forest, RFM Clustering, Natural Language Processing, Prompt Engineering, FAISS
• Data & Analytics: MySQL, SQLite, Pandas, Seaborn, Matplotlib, Power BI

📫 Connect with me: LinkedIn | findsamirks@gmail.com