Urv Patel

AI Engineer

Bimage • Período integral

Jan 2025 - Feb 2026 • 1 yr 1 mo

• Built an AI-powered research engine for a construction firm that cut manual engineering research time by 70%, saving the team over 30 hours every week by instantly surfacing answers from decades of compliance documents, tender files, and technical specifications. • Introduced a quality benchmarking system for all AI solutions before client deployment, ensuring every product shipped met defined performance standards — eliminating guesswork and reducing post-launch issues across the portfolio. • Fine-tuned a lightweight AI model on proprietary construction data, producing a domain-specific solution that significantly outperformed off-the-shelf models in understanding technical terminology and reduced inaccurate outputs in client-facing applications. • Built an autonomous documentation system that kept internal technical docs up to date in real time, cutting onboarding time for new engineers and external partners and removing a recurring bottleneck that previously required constant manual effort. • Delivered 3 full-stack internal products in weeks rather than months by combining AI-assisted development with n8n automation — including a client onboarding system that reduced manual intake from multiple days to under 2 hours.

Data Analyst

Consulting • Período integral

Jun 2023 - Dec 2024 • 1 yr 6 mos

• Engineered automated reporting pipelines using SQL, Power Query, and Python, eliminating manual extraction and reducing weekly reporting time by 60% to accelerate leadership decision-making. • Developed predictive models like regression and time-series in Python to forecast project completion and budget deviations from early-phase data, increasing long-term forecasting accuracy by 25%. • Designed dynamic Power BI and Tableau dashboards utilized by global project leadership to monitor schedule variance across a portfolio of 20 projects, directly driving an 18% improvement in on-time delivery. • Executed statistical and trend analysis on historical delivery data across 3 global regions, identifying operational bottlenecks and implementing process optimizations that reduced rework by 22%.