Featured Projects & Dashboards

Project Overview

An end-to-end e-commerce data pipeline that turns raw, unsorted product photography into ready-to-list eBay inventory with no manual data entry involved. Photos are batched, segmented, and sent through a serverless AI service that writes the SEO title, condition, description, and pricing for every item automatically.

Behind the Pipeline

Photos sync to a Google Cloud Storage bucket and are picked up every minute by a Python service running on Google Cloud Run. Items are separated automatically by detecting a "blackout" divider photo (the lens covered by hand between items) using image-brightness detection. Each item's photos are sent to the Gemini API, which returns a structured listing (title, condition, description, and estimated value/shipping) as clean JSON. The finished listings are then loaded into a Databricks SQL warehouse, managed day-to-day through a custom Streamlit terminal, and tracked in Power BI to compare AI-estimated pricing against what items actually sell for.

Key Highlights

• Physical Segmentation, No Manual Sorting: A single blackout photo, lens covered by hand, is enough to split a 200+ photo batch into individual items, with no folder organizing or manual tagging required.
• AI-Generated Listings: The Gemini API reads each item's photos and produces a ready-to-post eBay listing (title, condition, description, pricing) in one structured pass.
• Closed-Loop Analytics: Power BI compares the AI's estimated value against the actual sold price, turning every completed sale into feedback on how well the model is pricing inventory.

Read the Full Case Study →

Project Overview

This dashboard provides a comprehensive visual analysis of over 8.3 million consumer complaints filed with the Consumer Financial Protection Bureau (CFPB) between January 2025 and July 2026. By structuring and visualizing this massive dataset, the report identifies systemic issues, geographic pain points, and company-specific trends within the financial sector.

Behind the Dashboard

This isn't a one-time export — the report is wired directly into a live data pipeline. New complaint data is pulled from a Databricks workspace via a scheduled API call, then automatically transformed and loaded into the reporting layer. That means the numbers above refresh on their own, with no manual downloads or file handling required to keep the dashboard current.

Key Insights Discovered

• Credit Bureaus Dominate the Data: The vast majority of consumer grievances are directed at the "Big Three" credit reporting agencies. TransUnion, Equifax, and Experian account for over 7.1 million of the total complaints analyzed.
• Root Cause Identification: Utilizing decomposition tree analysis, the data reveals that "Credit reporting" is the primary product of concern. Drilling down further, "Incorrect information" acts as the leading sub-issue, driving over 4.3 million specific complaints.
• Volume Trends & Geography: The 3-month rolling average highlights a steady, continuous upward trend in complaint volume from early 2025 through mid-2026. Additionally, geospatial tracking maps out regional densities of delayed company responses (taking over 3 days).

Read the Full Case Study →

Project Overview

This report analyzes 40,529 public CFPB complaints about Chase from January 2025 to July 2026. I grouped the CFPB's 75 issue labels (68 distinct issues once I merged the ones the CFPB renamed mid-period) into 10 customer-journey friction themes, then benchmarked how complaints get resolved against five peer banks. The goal is to find where customers hit friction, size it, and point to what a team could test first.

Behind the Dashboard

Complaints were pulled month by month from the CFPB API, then cleaned and normalized in Python (deduplicating, harmonizing issue names the CFPB changed mid-period, and mapping states to regions). Friction themes and customer segments were engineered in pandas, and the Power BI model was built as a version-controlled project, with every measure validated against the Python numbers before publishing.

Key Insights Discovered

• Money Problems Push Customers to the Phone: Complaints about getting access to funds, the largest friction theme at 9,842 complaints, are filed by phone at 5.0%, twice the 2.5% average.
• Disputes Are Growing Fast: Dispute and investigation complaints are up 55.8% year over year (January to July), a likely driver of "where is my dispute" status contacts and a natural candidate for proactive updates.
• Resolution Lags Peers: Chase closes 14.4% of complaints with relief versus a 27.4% peer average, and the gap holds within the same product. It is a hypothesis to test, not a verdict, since relief is the company's own closure code.

Note: this is public complaint data, not call-center telephony. It shows where friction escalates, not call volume or containment.

Project Overview

This dashboard tracks Service-Level Agreement (SLA) performance across a contact center's 52,000+ customer calls in 2021. It gives leadership a clear, day-by-day view of how consistently the team is meeting its response and handling targets, and points to where call length is putting that performance at risk.

Behind the Dashboard

Unlike the CFPB report above, which pulls automatically from a cloud data platform, this pipeline runs entirely locally, end to end. Raw call records are extracted, cleaned, and loaded using a custom Python-built ELT process into a local database, then modeled and published into Power BI. Without a managed cloud platform doing the heavy lifting, this required hands-on engineering at every stage: scripting the extraction and transformation logic, managing the local database, and scheduling the refresh process — the kind of ground-up pipeline work that goes into building reliable reporting from scratch.

Key Insights Discovered

• SLA Performance is Strong, but There's a Gap: 92% of all calls met the Service-Level Agreement target in 2021. The remaining 8% — roughly 4,100 calls — represents missed commitments worth tracking closely, since each one is a customer who didn't get the response they were promised.
• Wait Times Aren't the Bottleneck — Handling Time Is: Customers wait only 17 seconds on average before reaching an agent, but the call itself takes nearly 5 minutes to resolve. Since handling time makes up almost the entire length of a call, it's the biggest opportunity for improving both efficiency and customer experience.
• Longer Calls Drive SLA Misses: Days with longer average handle times consistently show lower SLA compliance, with some dropping into the 80% range once handling time crosses roughly 5 minutes. This points to a clear opportunity: shortening time spent per call, through agent coaching, better tools, or streamlined processes, is likely the fastest way to lift SLA performance above the current 92% baseline.

Read the Full Case Study →

Project Overview

A multivariate exploration of the Airline Passenger Satisfaction dataset (Kaggle), covering 103,904 passenger records across 23 variables. The analysis applies a lens shaped by years of contact center CX work, loyal customers vs. one-and-done customers, what actually separates them, testing service ratings, trip context, and delays against loyalty to see which dimensions actually move the needle.

Behind the Analysis

Built entirely in a Jupyter notebook with pandas, matplotlib, and seaborn for exploration, the analysis works through the data in layers rather than jumping straight to a model: first checking whether loyalty simply tracks who's flying, then ranking all 14 service-rating dimensions by the gap between loyal and disloyal customers, then testing operational reliability, before finally training a Random Forest classifier to weigh every factor together. The model reached 94.6% accuracy separating loyal from disloyal passengers.

Key Insights Discovered

• Who's Flying Beats How the Flight Went: Trip context, mainly business vs. personal travel, age, and trip length, outweighs the entire service experience in explaining loyalty.
• Booking and Boarding Stand Out: Among 14 service-rating dimensions, online boarding, on-time convenience, and easy booking separate loyal from disloyal customers the most. Comfort and in-flight extras barely register.
• Delays Don't Move Loyalty: Delay frequency and severity are nearly identical between loyal and disloyal passengers, the "a late flight kills loyalty" assumption doesn't hold up in this data.

Read the Full Case Study →