AI Pipeline
Flask
Claude API
01
Lead Scout
Built by an AI adoption trainer who got tired of searching for clients manually.
Before
Manually searching LinkedIn and Google for companies that might need AI training — hours of tab-switching with no consistent scoring or output.
Build
A five-stage pipeline: discover → score → classify → find contact → draft email. Results stream live to a UI as each company clears the pipeline.
Result
Select an industry and region, click Scout — get scored leads with a named contact and a personalised cold email draft, in minutes.
Python
Flask
Anthropic API
Tavily API
Hunter.io
BeautifulSoup
SSE
Bellabeat Case Study
The data contradicted the hypothesis — and that turned out to be the finding.
Before
Bellabeat lacked a data-driven argument for which product feature to build next. Competitor feature comparisons existed, but no user behaviour data to validate the priority.
Build
End-to-end analysis of 30 Fitbit users' activity, sleep, and heart rate data — cleaned in SQL, feature-engineered in R, visualised in Tableau. Google Data Analytics capstone.
Result
Users slept poorly and then worked out harder — not less. That compensation pattern was the finding. It invalidated the performance feature on the roadmap and validated a Burnout Prevention feature instead. One dataset. One redirected product decision.
R / tidyverse
SQL (BigQuery)
Tableau
Google Sheets
Scraping
Data Analysis
03
Developer Portfolio Scanner
Recruiters spend hours manually browsing developer sites. This automates that.
Before
A recruiter looking for available developers visits portfolio sites one by one — checking for availability signals, inferring skills, copy-pasting into a spreadsheet. Slow, inconsistent, and manual.
Build
A modular Python scraper that detects availability signals and extracts skills across 25 tech categories using regex. v2 added Pandas cleaning, score normalisation, and PNG chart output.
Result
Give it a list of URLs and get back a clean, scored dataset of developer availability and skills — plus visualisation charts — in seconds instead of hours.
Python
BeautifulSoup
Requests
Pandas
Matplotlib
AI Pipeline
Data Analysis
Claude API
04
Sales Performance Analyzer
Most companies have more sales data than time to look at it.
Before
A sales manager ends the week with a CSV export and a Monday meeting to prepare for. The analysis that should take an afternoon gets compressed into twenty minutes of gut-feel summary.
Build
A Python pipeline that loads the CSV, runs Pandas analysis across categories, regions, and monthly trends, then sends a structured summary to Claude for a plain-English executive report.
Result
A formatted report with revenue breakdown, trend analysis, anomaly flagging, and three concrete recommendations — generated in seconds, saved to disk.
Python
Pandas
Anthropic SDK
python-dotenv