I'm a Data / Business Intelligence Analyst who builds end-to-end analytics solutions — from cloud data pipelines and SQL modeling through to interactive dashboards that drive decisions. I design star schemas, write performance-minded DAX and SQL, and turn large, messy datasets into reporting that's clean, trustworthy, and business-ready.

My portfolio spans cloud data engineering on both Azure (incremental ELT with Data Factory, a Databricks lakehouse, SCD Type 2 dimensions) and Microsoft Fabric (metadata-driven pipelines, T-SQL warehouse transformations, semantic models), alongside Power BI & SQL (star-schema semantic models, performance-minded DAX), PostgreSQL & Python (a fraud-analytics app on a 6.3M+ row dataset), and Excel FP&A (Power Query → Power Pivot → DAX). I care about the unglamorous parts that make BI trustworthy — idempotent ETL, grain integrity, and numbers that reconcile end to end.

An end-to-end Azure data platform: watermark-driven incremental extraction from Azure SQL into a medallion lakehouse on ADLS Gen2, governed by Unity Catalog with managed-identity access — no keys or SAS tokens in code. A single metadata-driven ADF pipeline loads five tables by iterating a JSON parameter array, with a conditional branch that cleans up the empty files ADF writes on no-op runs. Auto Loader streams bronze → silver; Delta Live Tables builds a star schema with SCD Type 2 dimensions so a stream event joins to whichever subscription tier a user held at the time. Validated rather than assumed: 509 gold rows resolving to 500 current + 9 superseded versions, with an integrity check confirming exactly one open row per key.
Azure Data Factory · Databricks · Delta Live Tables · Unity Catalog · PySpark · Delta Lake · SCD Type 2

A governed analytics platform for a city transportation authority, built entirely on Microsoft Fabric: monthly parquet files land in a Lakehouse, metadata-driven pipelines load a Warehouse staging schema one month at a time, and T-SQL stored procedures transform and append into a historical presentation table feeding a semantic model and Power BI report. A processing_log watermark makes each run self-managing — no parameters to edit, no double-processing. The transformation was built both as a Dataflow Gen2 and as a stored procedure, and the measured comparison drove the final design: presentation processing dropped from ~2–3 minutes to ~30 seconds.
Microsoft Fabric · OneLake · T-SQL · Data Pipelines · Semantic Model · Power BI

A BI dashboard on 6.3M+ PaySim transactions, built around one mandate: cut fraud-investigation false positives without losing catch rate. The headline finding — a single high-precision rule catches 76% of all fraud at 97% precision, while a second noisy signal inflates flagged volume to 2.5M at near-zero precision — became a concrete recommendation to drop the diluting signal. Live dashboard ↗
Python · PostgreSQL · Streamlit · Data Modeling

An end-to-end BI case study for a mid-size omnichannel retailer: raw sales, inventory, and movement data modeled through DuckDB + Parquet + SQL into a Power BI executive dashboard with a Kimball-style star schema and a leadership-ready findings deck. The analysis resolved an operations-vs-finance dispute by proving the real risk was overstock, not stockouts — ~$95K of working capital trapped in slow movers (24 of 40 SKUs over 90 days of cover), framed into a 90-day rebalancing plan worth a $50–70K recovery opportunity.
Python · DuckDB · Parquet · SQL · Power BI · DAX

A one-click FP&A reporting dashboard in Excel: change a single date cell, hit Refresh All, and every KPI, variance, and chart updates. Power Query unpivots monthly P&L grids into tidy fact tables; a Power Pivot star schema (actuals + budget facts over shared date / account / department dimensions) drives DAX time-intelligence measures — current month, trailing 3 / 6 / 12, and YTD with prior-period and prior-year comparisons. Surfaces budget-vs-actual variance, gross-margin bridges, and operating-expense breakdowns by department and cost type, with KPI tiles and conditional gauges. Includes technical docs and a CFO-style findings memo.
Excel · Power Query · Power Pivot · DAX · FP&A · Budget-vs-Actual

Two linked Tableau dashboards (Sales + Customer) on 9,994 retail transactions (2020–2023), tied together by a shared Select Year parameter that drives every year-over-year comparison. Built around real findings, not just charts: a discount margin cliff at 20% (every tier above it is unprofitable, down to −122.6%), profit concentration far sharper than 80/20 (8.8% of products drive 80% of profit, while 301 products quietly destroy $76.7K in margin), and a 57.7% single-purchase customer tail. Published to Tableau Public.
Tableau · Parameters · YoY Analysis · Profitability · Customer Analytics

📌 More case studies and write-ups are in my pinned repositories below.

• Extending the pipeline stack beyond a single cloud — an orchestration-focused build pairing Apache Airflow (Docker) with dbt Core on Databricks, covering DAG design, scheduling, dependency management, and tested modular transformations
• Polishing recruiter-ready BI case studies end to end: SQL model → DAX → dashboard → documentation
• Preparing for DP-700: Microsoft Fabric Data Engineer Associate

Open to roles in Data Engineering, Analytics Engineering, Business Intelligence, and Data Warehousing — where I can contribute through pipeline development, dimensional modeling, SQL analysis, and reporting people can trust.