DPF Labs
Daniel P Francisco
LAIFT President · Pharmacy student
Pharmacy student and LAIFT President. I build AI-powered health education platforms, somewhere between the laboratory bench and the code.
Read moreShow less
Pharmacy undergraduate at UNINASSAU Salvador, focused on compounding pharmacy and pharmaceutical technology, with an interest in digital health and telepharmacy.
President of LAIFT (Interdisciplinary Academic League of Pharmacology and Toxicology), where he leads teams and designs HealthTech teaching solutions to democratize hands-on learning.
Uses generative AI and prompt engineering to structure knowledge and build simulators, interactive systems and an integrated learning platform.
Brings prior experience in customer service, operational risk management (NR-20) and strict execution of standard operating procedures (SOPs), skills that carry over to laboratories and clinical studies.
Journey
- 2026 – present
President
LAIFT · UNINASSAU
Strategic management of the league, leading teams in spreading interdisciplinary knowledge in pharmacology and toxicology. Design and development of HealthTech teaching solutions with AI and digital tools, published on GitHub.
- 2025 – present
Customer Relationship Specialist I
Atento Brasil
Customer service and relationship management, with active listening and assertive communication.
- 2024
Fueling Operator and Risk Management
Ferreira Combustíveis
Handling and logistics of flammable and hazardous products, following NR-20, SOPs, supply inventory control and fire and chemical accident protocols.
- 2023 – 2024
Seaman
Brazilian Navy · 2nd Naval District Command
Strict execution of standard operating procedures, hierarchy, discipline and readiness in a high-complexity environment.
Education
- Bachelor of Pharmacy (in progress)
UNINASSAU Salvador · 2024 – 2029 - High school
Colégio Estadual José Ribeiro Pamponet · 2020 – 2022
Certifications
- Patient Safety and Quality in Health Services (100 h) · ENAP
- Total Quality Management (15 h)
- Injectable Administration
- Dangerous Goods Regulations
Languages
- Portuguese · Fluent
- English · Basic
Technical output
- Interactive Pharmacological Assessment and Learning System (Gamified Interface) (2026)
- Interactive Chemical Reaction and Flame Spectroscopy Simulator (2026)
- Structured Database for Curation of Reference Pharmaceutical and Medical Literature (2026)
Focus and interests
- Compounding pharmacy and pharmaceutical technology
- AI, digital health and telepharmacy
- Clinical and translational research: treatment adherence and clinical guidelines (PCDT) for neglected diseases
AI and digital health
Generative AI and prompt engineering to structure hypotheses, curate content and build health education solutions.
Pharmacy and science
Pharmacology and toxicology applied to teaching; study focus on pharmaceutical technology and analytical chemistry for compounding pharmacy.
Quality and safety
Patient safety, quality management and rigor in executing SOPs, with a routine of safety standards in high-risk environments.
Leadership and management
Team and project coordination, organization of academic routines and clear communication learned in customer service.
Behind the scenes
See all8 modules
Modules
Each module is a standalone teaching tool, integrated into the same platform.
Plataforma LAIFT
The platform
What was designed and built so that an AI-powered health education platform is useful, safe and trustworthy.
Plataforma LAIFT
LAIFT unified platform: community governance (events, proposals with voting, tasks, team, messaging) and the "Learn" ecosystem (pharmacology and toxicology simulators, virtual clinic, laboratory and 3D anatomy atlas), with built-in artificial intelligence.
Want to see it working? Visit the platform or ask for a demo.
Extra curiosity
Part of the platform code is public. It is not the main way to get to know the project, but the invitation is open to the curious.
View the code on GitHubArchitecture
- 1Lightweight interfaceFast web interface with no dependence on large frameworks, working on desktop and mobile. Learning modules share the same session, with no extra sign-in.
- 2Edge APIService layer that only runs pre-approved actions and checks identity and permission on every call, close to the user for low latency.
- 3Relational databaseStructured data with parameterized queries, versioned migrations, consent history and concurrency control for one-time actions such as voting, seats and check-in.
- 4AI layerLanguage models serving teaching: virtual patient, preceptor, case generation and a platform assistant, with quotas, moderation and human review of what becomes shared content.
Profiles and permissions
Who can do what, designed so that power and accountability go together.
- ProblemImproper access to management functionsSolutionThree base roles (visitor, member and administrator) combined with functional roles: manager, content reviewer, moderator, curator and mentor.
- ProblemRoles with too much powerSolutionCapability-based permissions, not title-based: each role unlocks only what the function needs, and several roles can be combined.
- ProblemCompromised management accountsSolutionMandatory two-step verification for roles with power over people or content.
- ProblemIdentity forged in the browserSolutionIdentity, role and account status verified on the server on every action, never trusted from the client.
- ProblemPower granted without a traceSolutionRole assignment restricted to administration, with an audit trail for sensitive actions.
Security and privacy
Layered protection, from the browser to the data, with privacy as a design requirement.
- ProblemMalicious content typed by someoneSolutionAll text coming from people or AI is treated as plain text, and the browser receives a restrictive content policy.
- ProblemAutomated access attemptsSolutionAttempt limits on sign-in, sign-up and password recovery, with anti-bot protection where use without an account is allowed.
- ProblemPassword leaksSolutionPasswords stored only as cryptographic hashes; resetting a password ends all open sessions.
- ProblemUse of data without consentSolutionVersioned terms of use and privacy policy, with acceptance of each version recorded and preserved.
- ProblemExposed conversationsSolutionEnd-to-end encrypted messages between league members.
- ProblemExported spreadsheets with dangerous contentSolutionExports protected against formula injection.
Responsible AI
AI as a teaching aid, with limits on cost, content and privacy.
- ProblemCost and misuseSolutionDaily quotas per person and per feature, with a global limit and automatic cut-off on failure.
- ProblemInappropriate contentSolutionModeration of questions and answers, with a human review queue for anything flagged.
- ProblemUnfounded answersSolutionCurated content, cited sources and case generation with defined profiles and assessment criteria.
- ProblemMalicious instructions hidden in textSolutionSeparation between user-written content and system instructions; nothing becomes shared content without review.
- ProblemExposure of personal dataSolutionQuestions logged in a minimized form without storing the raw text, and a clear notice about what is sent to the AI provider.
Learning, management and quality
From individual study to league management, with editorial and technical care.
- ProblemUnreliable anatomy contentSolutionPortuguese content sheets with human editorial review before publication in the atlas.
- ProblemStudy without follow-upSolutionStudy statistics, study memory, a tutor and performance-axis assessment in the virtual clinic.
- ProblemAttendance fraud at eventsSolutionSigned QR Code credential and a check-in terminal with permission-restricted actions.
- ProblemLeague decisions with no recordSolutionProposals with voting, events, collaborative tasks and an org chart inside the platform.
- ProblemRegressions and inaccessibilitySolutionAutomated tests, screen-reader accessibility checks and a staging environment separate from production.
Evolving
Directions the platform may take. The tag shows where each one stands; none is a promise with a deadline.
- In testing
Tutor with study memory
A tutor that remembers what a person has studied, only if they allow it, and adapts explanations and reviews. The memory is kept apart from other data and can be revoked or erased at any time, and never stores personal or health data.
- In testing
AI personas with safeguards
Assistants with defined roles (tutor, preceptor, advisor) and input and output rules: they refuse out-of-scope requests, never quote doses and leave a usage trail without storing the person's text.
- Under evaluation
An assistant that reasons better over the library
More precise selection of the passages that answer each question, intent detection (is it health-related? does it need a source? is it in the league's scope?) and continuous evaluation of answer quality.
- Under evaluation
Assisted scientific search
Search across open literature databases with a short summary and the source always in view. Anything stating practice, dose or interaction goes through human clinical review before it appears.
- Under evaluation
Growing scientific library
More articles in Portuguese, curation by area with public inclusion criteria, themed collections, favorites and citation in ABNT and Vancouver formats.
- In testing
Atlas with a systems view
Body navigation by system (cardiovascular, nervous, digestive and others), new content batches reviewed by the editorial board and integration with pharmacokinetics.
- Planned idea
Applications and selection inside the platform
Call for applications, sign-up, follow-up and results in one place, with openings and dates published from a single source.
- Under evaluation
Sign-in with passkeys
Simpler sign-in that is harder to phish, using the person's own device, for when the platform handles more sensitive data.
- Planned idea
Partnerships and integration with other institutions
Token-controlled access for those who want to consume content and aggregated data, and versions in other languages for leagues and students abroad.
A description of what was designed and implemented on the platform. It is not a certification or an external audit.
5 photos
Behind the scenes
Laboratory, classroom and the way here, in a relaxed tone.
Salvador, Bahia, Brazil
Contact
Interested in talking about digital health, AI in education or research? Get in touch.