Personal details
Charles F.
About
Applied AI Engineer with 13+ years of software engineering experience and an MSc in AI/ML. I build production LLM systems — agent platforms, document-understanding and retrieval pipelines, and the backends that run them — with a focus on making AI reliable, deployable, and tied to measurable business value.
Most recently I've been the backend / AI-platform engineer on Maestro, a venture-backed mortgage-origination platform, building the multi-tenant, agent-facing loan-record service that every AI agent reads and writes through — with database-enforced tenant isolation, optimistic concurrency, append-only audit logging, and field-level PII encryption. Alongside client work I build and ship my own AI tools, including a document-exploration agent that uses sandboxed code execution as an alternative to RAG, and DSPy-based LLM pipelines and evaluations.
My foundation is a deep mobile background — 13 years of iOS/Swift, including scaling a platform to 1M+ users and shipping apps with exceptional metrics (17-minute average sessions, 4.9 App Store ratings). That product-and-reliability mindset is exactly what I bring to AI: systems that hold up in production, not just demos.
Work Experience
- Building the canonical loan-record service at the center of an AI-agent-driven mortgage-origination platform — a multi-tenant REST API that serves as the single source of truth every AI agent reads and writes through (never directly to the database), accessed via an MCP tool layer I co-designed.
- Architected the service to the platform's target scale: ~250k new loans/year with tens of thousands of concurrently active loans, each touched by 10+ AI agents over a multi-week lifecycle.
- Designed and implemented database-enforced, fail-closed multi-tenant isolation (PostgreSQL row-level security) so cross-tenant data access is structurally impossible — suitable for regulated financial data containing PII.
- Built the section-level update API with optimistic concurrency (ETag/version-based), atomic multi-section writes, and an append-only, immutable audit log capturing before/after state for every change.
- Implemented envelope encryption for sensitive borrower PII (e.g., SSN, DOB) backed by a managed key service, with audit-log redaction so secrets never persist in history.
- Own the workflow/task engine tracking each loan's lifecycle, including blocking "gate" tasks that pause an orchestrated agent workflow until resolved; designed and built the service's CI/CD pipelines on GitHub Actions, driven by strict TDD against real database behavior.
AI Tooling & Agentic Engineering:
- Build and ship independent AI tools — including a document-exploration agent that uses sandboxed Python execution as an alternative to RAG (live on appsimple.io), and DSPy-based LLM pipelines, evaluations, and prompt-optimization workflows.
- Direct large, multi-language systems built primarily through AI coding agents, using deliberate manager–worker and parallel-worktree patterns rather than ad-hoc prompting.
Applied AI & Automation:
- Design and implement practical AI solutions to enhance workflows, automate processes, and unlock business value.
- Build data parsing and transformation pipelines that extract, validate, and structure data from multiple sources (APIs, Excel, JSON, unstructured, etc.) for reporting and visualization systems.
- Develop AI strategies, evaluate model performance, and build intelligent document processing solutions.
- Focus on creating AI applications that are reliable, deployable, and deliver measurable results.
Expert Mobile Development:
- Build high-quality, user-centric native iOS and Android applications, drawing on extensive experience shipping apps used by hundreds of thousands.
- Develop cross-platform solutions and manage the full mobile product lifecycle from concept to launch and iteration.
- Specialize in creating engaging mobile experiences, content delivery platforms, and subscription-based applications.
Hybrid Solutions:
- Architect and develop integrated systems that combine intuitive mobile interfaces with powerful AI capabilities.
My approach is grounded in pragmatic engineering principles, ensuring that whether the solution involves AI, mobile, or both, it is built for scalability, usability, and real-world impact.