Data Analyst Uncovering the hidden patterns that turn guesses into growth. 📍 Based in Sylhet, Bangladesh

• SQL — Joins, CTEs, Window Functions, Aggregations
• Power BI — Data Modeling, DAX, Calculated Columns, Power Query, Interactive Dashboards
• Python — Pandas, NumPy
• Excel — Pivot Tables, Power Query, Advanced Formulas, Charts
• Core Analytics — Data Cleaning, EDA, Business Intelligence, Data Storytelling
• Version Control — Git & GitHub

Technologies: MySQL, Power BI (DAX, Data Modeling, Calculated Columns, Power Query)

Analyzed ~829K transaction records across 50 stores in Mexico to surface actionable business insights:

• Diagnosed a -11.3% gross margin decline despite 30.9% revenue growth and 40.8% increase in units sold
• Identified Electronics as the highest margin category (44.6%) yet underleveraged at only 15.55% revenue share
• Flagged 20 SKUs at active stockout risk across all locations, including top seller Lego Bricks with 3 days of stock left
• Built a 5-page interactive Power BI dashboard with drill-through to individual store level across all 50 locations

→ View Repository

Technologies: Databricks (PySpark, SQL), n8n, PaySim Dataset, Telegram Bot API

Built a cloud-scale fintech pipeline processing 6.3M mobile money transactions end to end, from raw ingestion to automated business alerts:

• Modeled a star schema in Databricks with fact/dimension tables and daily account balance snapshots
• Computed daily and weekly KPIs including fraud flag rate, transaction volume, and type mix
• Designed 4 explainable, rule-based risk triggers (transaction spikes, transfer-to-cashout chains, fraud rate breaches, volume collapse), validated against real dataset outputs rather than a black-box model
• Orchestrated 3 cron-scheduled n8n workflows that query Databricks directly and push formatted daily/weekly summaries and emergency alerts to Telegram

→ View Repository

Technologies: n8n, Google Drive, Pinecone, Google Gemini

Built a fully automated retrieval-augmented generation system in n8n that watches a Google Drive folder and lets you chat with an AI agent grounded strictly in that knowledge base:

• Designed dual Google Drive triggers (create + update) to reliably catch every document change, chunked and embedded new content into Pinecone with no manual re-run required
• Built a query pipeline where the AI agent retrieves from the vector store before answering and explicitly declines when the knowledge base lacks the answer, avoiding hallucination
• Verified live knowledge base updates end to end: added a new document mid-session and confirmed the chatbot could answer questions only that document contained

→ View Repository

• Advanced SQL (Complex Window Functions, Query Optimization)
• Python for Data Analysis (Matplotlib, Seaborn)
• Workflow Automation

• Email: pallabdey21@gmail.com
• Portfolio: pallabdey.com
• LinkedIn: linkedin.com/in/pallabdey007
• GitHub: github.com/Rusher0077

"In God we trust; all others rely on numbers 😄"