Internal R&D · Azure AI
GovCon Opportunity Intelligence: AI bid triage on Azure
We built our own Azure-native platform that reads new government solicitations across 10+ monitored public sources — SAM.gov, USASpending, Grants.gov and FPDS among them — and turns them into a ranked, explainable bid pipeline: AI-scored fit with hard eligibility gates, certification-gap analysis, and an agent-generated requirement-by-requirement compliance matrix for every shortlisted opportunity.
Challenge
Opportunities scatter across dozens of portals in inconsistent formats. Manual triage misses openings and burns senior time on unwinnable bids — and any automation must collect only public information, with no access to nonpublic source-selection material, ever.
Solution
Serverless Azure platform from Bicep infrastructure-as-code: Functions-orchestrated ingestion, embeddings in Azure SQL native vector search, transparent two-tier AI scoring with cost governance enforced in code, and a Microsoft Foundry agent that reads solicitation documents and returns a strict, auditable JSON compliance matrix.
Outcome
New solicitations from every monitored source triaged automatically into a live pipeline; analysts start from a ranked shortlist with eligibility, certification-gap and incumbency pictures already assembled. Public-data-only collection engineered at four layers — source allowlist in code, database constraints, robots.txt-honoring collectors, AI prompt policy — including declining a high-value source whose terms disallow collection.
Proof of rigor: ~9,200 lines of Python, Bicep, TypeScript and SQL · 93 backend tests plus web and infrastructure test suites · 14 Bicep modules · keyless, managed-identity security end to end.
Azure OpenAI · Microsoft Foundry agents · Azure Document Intelligence · Azure Functions · Azure SQL vector search · Container Apps Jobs · Static Web Apps + React · Entra ID · Key Vault · Bicep IaC
Client delivery · Municipal water sector
AI-driven pricing and proposal automation for municipal water infrastructure
A US supplier of water and wastewater treatment equipment priced municipal projects by hand. We digitized ~28 historical offers spanning roughly 14 US states into regression-based pricing models and delivered a secure Power Apps application on Microsoft Dataverse that calculates a fully costed, tariff-aware quote in seconds — and generates the complete branded proposal document in one click.
Challenge
Pricing for municipal treatment equipment lived in spreadsheets and PDF offers; a multi-step cost chain — currency, freight, tariffs, overhead, margin — was manual and error-prone, with no audit trail from issued proposals back to inputs.
Solution
Custom Dataverse data model built as repeatable code under our own solution publisher; multilinear-regression pricing engine (R² up to 0.999, mean error as low as 0.7% on historical offer data) with floor-price guards and manual override; one-click Word proposal generation; least-privilege security roles; Entra ID authentication.
Outcome
Historical offers became a living, data-driven pricing capability. Quotes moved from spreadsheet work to instant, policy-consistent calculation; every proposal is versioned with a governed Draft → Calculated → Finalized workflow and a complete audit record for every generated document.
Microsoft Dataverse · Power Apps Code Apps · Power Platform ALM · Microsoft Entra ID · React · TypeScript · client-side document generation
Client delivery · Fortune 50 energy
An initiative portfolio that continuously creates value and reduces cost
For a Fortune 50 energy company, we rebuilt project-portfolio management on the Power Platform — demand intake, like-for-like comparison, staged rollout — with SharePoint, Power Automate and Power BI carrying the workflow.
- 40% fewer compliance exceptions across the governed portfolio
- 30% faster delivery schedules across portfolios
- 20 hours per month returned from financial reporting
Download the three-page case study (PDF)
Power Platform PPM · SharePoint · Power Automate · Power BI
Client delivery · Applied AI
Finding the right target, before anyone else
For a manufacturing sales organization, an AI vision pipeline over satellite and GIS imagery combined with public EPA and BLS data ranks business-development targets — compressing lead-to-outreach from 15–20 days to 3–5 and lifting conversion above 75%. These are the client's own before-and-after figures, and are directional.
Download the three-page case study (PDF)
OpenAI vision models · GIS imagery · EPA and BLS public data · AI agents
Client delivery · Regulated compliance
Compliant approvals in a day — validated to 21 CFR Part 11
A global pharmaceutical company — 37 sites, 2M+ pages digitized — needed compliant approvals at scale. We delivered electronic-signature workflows on SharePoint with multi-factor authentication and automated routing, validated to 21 CFR Part 11 across three environments.
- Signature cycles from 8 days to 1
- 75% first-pass success in FDA mock audit
- 85% shorter approval lead times; 25% fewer batch-release deviations
Why it matters to government: validated-environment delivery — audit trails, controlled access, data integrity — is the same discipline federal records and compliance programs demand.
SharePoint · DocuSign · Entra ID MFA · Power Automate · validated environments
Client delivery · Public-sector adjacent
Automating COVID-era housing assistance at scale
For a nonprofit administering emergency housing assistance, we automated intake and case processing during the COVID-19 response — public-program delivery under time pressure, with auditability the funders could trust.
Microsoft Power Platform · document automation · workflow governance
References, scope confirmations and points of contact for the client engagements above are available to contracting officers and teaming partners on request — written client permission governs what we name publicly.