Harsh Patel

The founding team was making product and go-to-market calls based on scattered notes, half-finished spreadsheets, and whatever someone remembered from a demo last week. No shared view of who was doing what.

Pulled together a structured competitive view of 20+ construction tech companies, tracking pricing, product features, positioning, and recent moves like funding or acquisitions. Pulled signals from company websites, review sites, and public sources, and organized everything into a consistent shape so competitors could actually be compared side by side.

Manually reading through dozens of competitor sites and pricing pages was eating most of the week, and different research rounds kept producing slightly different shapes of data.

Designed prompt-based research workflows that used LLMs to extract structured info from messy web sources: pricing, feature lists, target customers, and positioning language. Standardized the output format so each competitor scan slotted into the same template, and documented the prompts so the process stayed repeatable instead of one-off.

As a small team, "go after everyone in construction" was not a strategy. The founders needed a defensible way to pick which segments to chase first without relying on gut feel alone.

Scored prospect segments across company size, tech maturity, and pain severity to surface 3 priority verticals for early outbound. Then built a Power BI market intelligence dashboard that brought competitor feature releases, pricing shifts, and acquisition activity into one place, so competitive monitoring stopped being a weekly scramble.

Grant performance data over 8+ years lived in different spreadsheets and trackers. Every big review started with "which version of this is right?"

Helped build a reusable reporting model in Microsoft Fabric and Power BI so leaders could pull numbers from one place instead of chasing files. I worked on loading and cleaning historical spreadsheets, shaping them into a simple star schema for programs and projects, and running reconciliation checks so the new numbers matched what finance expected. I also documented how key metrics were calculated so different teams stopped maintaining their own formulas.

Leadership wanted a clearer, data-backed picture of Alberta's tech ecosystem to guide program focus, but information lived across reports, websites, and one-off discussions.

Pulled together public datasets, internal notes, and stakeholder input into a structured view of sectors, stages, regions, and funding gaps. Built summary tables and simple visuals that could drop into slide decks and briefings, and kept a small "sources & assumptions" log so people knew what the numbers actually represented.

Monthly funding summaries were built by copying numbers between spreadsheets. Any last-minute question meant starting another manual version.

Helped design a Power BI model on top of Fabric dataflows so core funding KPIs were always available in one dashboard. I contributed queries for program and project views, set up scheduled refreshes, and added simple validation checks so we could spot issues before meetings. For a few recurring asks, I used Python to tidy raw extracts into shapes the model could consume.

Several recurring reports were rebuilt by hand in spreadsheets every week, which left little time to discuss what the numbers meant.

Helped set up Power Automate flows and scheduled refreshes so key reports pulled their own data. Worked with my manager to test the new process against the old spreadsheets and fix mismatches before handing it over to the team.

Operational data came in with duplicates, inconsistent IDs, and missing values, which made simple questions hard to answer confidently.

Used SQL and Python (Pandas) to clean and reconcile 50K+ records into analysis-ready tables for leadership dashboards and ad-hoc questions. Logged the main assumptions and filters so future queries could follow the same rules.

Different teams reused the same metric names but calculated them slightly differently, which made cross-team dashboards confusing.

Worked with analysts and managers to write down 10+ KPI definitions, including formulas, filters, and caveats. Updated dashboard labels and notes so people could quickly see how a metric was built.