EarnedAI
Earned Value Management & Project Controls Intelligence Platform

I wrote the earned value calculations from first principles: PV, EV, AC, CV, SV, CPI, SPI, EAC (multiple methods), ETC, VAC, TCPI, and Earned Schedule (SV(t), SPI(t)). In a DOE oversight role, you need to know whether the contractor's numbers are correct — not just what a tool says. Building the engine myself means I understand every formula, every edge case, and every place a contractor might present misleading performance data.

Forward pass, backward pass, total float, critical path identification, and near-critical path flagging — all implemented from scratch. On DOE capital asset projects, schedule integrity is a gate review requirement. I built this because I've seen how easily logic errors, dangling activities, and broken predecessor chains slip through commercial tools. The cycle detection catches circular dependencies that would silently corrupt a schedule network.

All 14 checks implemented: missing logic, open ends, hard constraints, negative float, high float, high duration, invalid dates, resource loading, missed tasks, critical path percentage, critical path length index, BEI, lead/lag analysis, and relationship type distribution. These are the exact checks a federal oversight analyst runs during an IPS health assessment. I built the auditor to produce a composite quality score and specific remediation guidance — the same deliverable I'd expect to produce or review in a DOE PMSO role.

PERT and triangular distributions, configurable iterations, P10/P50/P80/P90 confidence intervals, criticality index per task, and sensitivity ranking. DOE CD-2 and CD-3 approvals require probabilistic cost and schedule range estimates. I built this to demonstrate that I don't just run Monte Carlo — I understand the statistical foundations, how distribution selection affects outcomes, and how to interpret criticality indices for schedule risk prioritization.

Rule-based engine that flags CPI/SPI degradation trends, cost-at-completion breaches, schedule variance inflection points, and WBS-level anomalies — each with a severity level, confidence score, and specific recommendation. This mirrors the kind of proactive surveillance a DOE project controls analyst performs: not waiting for a monthly CPR to reveal problems, but continuously scanning performance data for early indicators of cost overrun or schedule slip.

Custom parser for Oracle Primavera P6's proprietary XER export format, mapping activities, relationships, resources, and calendars into the application's data model. On every major DOE project, the Integrated Project Schedule lives in P6. I built the XER parser because real project controls work starts with real schedule data — not demo datasets — and oversight requires the ability to independently ingest and validate contractor-submitted schedules.

Aggregated BAC, EV, AC, CPI, and SPI across the full project portfolio with cross-project health comparison, variance ranking, and risk scoring. DOE site offices and PMSO teams don't oversee a single project — they oversee a portfolio. I built the aggregation layer to show that I understand how federal program managers need to see data: rolled up for executive reporting, but drillable to WBS-level detail for root cause analysis.

One-click executive narrative report generation that synthesizes EVM metrics, risk indicators, schedule status, and WBS variances into a structured 500-800 word analysis. In DOE oversight, the monthly project status report isn't just numbers — it's a narrative that explains what the data means, what the risks are, and what actions are recommended. I integrated Claude to demonstrate how AI can accelerate the reporting workflow while keeping the analyst in the loop for review and judgment.