Technical Product Lead & AI Solutions Architect
Turning Business Logic into Production Software
I combine 10+ years in the financial ecosystem (8 years leading direct credit operations) with 5+ years shipping production software and deterministic AI systems. Every architecture enforces statutory privacy (Law 172-13), explicit state machines, and strict TDD governance.
Legacy credit operations relied on fragmented spreadsheets, manual WhatsApp handoffs, and predictable database URLs that exposed borrower dossiers (IDOR risk).
I architected a decoupled credit engine in PHP 8 and MySQL paired with a sub-second Astro 5 intake portal. Every application moves through an immutable state machine with real-time KYC validation and SHA-256 cryptographic consent logs.
Astro 5 intake portal with regional geocontrol and persistent UTM attribution.
Dominican Modulo-10 ID checksum validation and real-time DGII employer tax verification.
Debt-service capacity modeling and Base62 tokenized re-engagement links (s.solufime.com).
Law 172-13 compliance with immutable SHA-256 signed consent payloads and audit trails.
Live operational queue filtering credit applications by deterministic state transitions and advisor SLAs.
RECIBIDA → VERIFICADA → ANALIZADA → APROBADA
Multi-agent orchestration, double-entry financial ledgers, and deterministic engineering governance.
Stateful LangGraph orchestration and PostgreSQL pgvector retrieval delivering <3s latency, 30% lower token costs via Redis semantic caching, and confidence-gated human handoffs in Chatwoot.
Core banking engine enforcing strict double-entry accounting invariants (Δ = 0.00), race-condition locking, and Four-Eyes Segregation of Duties across 43 Gherkin scenarios.
Monorepo combining FastAPI, Next.js, and the CreditGraph parser with shared schema contracts (frm-contracts) for automated risk profiling.
Multi-agent engineering protocol implementing the Uncle Bob 7-Agent Pipeline (Specifier → Coder → Refactorer → Architect → QA → PII-Verifier) with TDD and CRAP ≤ 6 limits.
Lead product architecture from domain discovery to production deployment in weeks rather than quarters.
Architect stateful LangGraph and pgvector RAG systems with strict confidence gates, PII scrubbing, and human escalation.
Enforce statutory privacy (Law 172-13) and accounting invariants, eliminating IDOR risks across sensitive credit workflows.
Before writing production software, I spent eight years managing direct credit operations—evaluating thousands of TransUnion and DataCrédito bureau reports, keeping historical direct-underwriting delinquency below 3%, and building risk matrices.
When I architect financial systems, I do not need an intermediary to translate underwriting rules, KYC, or audit constraints. I built them from the ground up.
COOPSOLUFIME • 2024 – Present
Elected by the General Assembly to direct internal oversight, liquidity audits, portfolio quality verification, and regulatory compliance under Ley 127-64.
Credit risk engines, DDD, double-entry ledgers, Modulo-10 checksums, Law 172-13 forensic privacy.
LangGraph stateful graphs, pgvector semantic RAG, Redis caching, Human-in-the-Loop handoffs.
Python 3.12+ (FastAPI, Pydantic), TypeScript, Astro, PHP 8, PostgreSQL 16, MySQL, Redis 7, Docker.
