Rafael Pinto
AI Business Analyst & Project Manager
From process to PoC, from PoC to adoption.
From problem to adoption.
Discovery
Process before technology. I scope from the very first contact with the client, gather needs, identify where value is being lost and estimate the effort.
Methods: stakeholder interviews, user stories and acceptance criteria, use cases, BPMN, wireframes, functional specifications.
Evaluation
I decide with data, not fashion. I define criteria and evals to choose models and libraries, balancing performance, available infrastructure and compliance.
Focus: LLMs/SLMs, ASR, NLP, on-premises inference, European Portuguese, GDPR and health data.
Building
I stay hands-on. I build PoCs to test ideas with clients early, before anyone invests. Several became sold projects and solutions in use.
Focus: generative AI, RAG, embeddings, AI agents, virtual assistants.
Delivery
I run two to three consulting projects at once, with backlogs and roadmaps that balance business value and effort. Every delivery goes through functional testing before it reaches the client.
Methods: Scrum, Kanban, backlog and roadmap management, team coordination.
Adoption
A solution is only worth something if it is used. I run AI assessments and training in companies, with diagnostics, adoption roadmaps and prioritised projects.
Focus: training in AI fundamentals and in tools such as Claude, ChatGPT and Microsoft 365 Copilot. Certified trainer (CCP).
Business Analyst
Process before technology. I scope from the very first contact with the client, gather needs, identify where value is being lost and estimate the effort.
Discovery: from first contact to proposal
These are the practices I use today. They are not a rigid process: I adjust them to each client, each project and what is already known about the process in question.
First contact (Client)
The client presents a need or an idea.
Follow-up (Me)
I follow up on the contact and capture the client’s idea.
Understand the need (Me)
I work out what the client needs, what kind of development could help and what needs lie behind the request.
Feasibility (Me)
I identify whether it is possible and form a macro view of how to execute it.
Validate with the team (Team)
I run validations with the team and explore different strategies and approaches.
Proposal (Me)
I consolidate the approaches into proposals, working out where there is value for the client and estimating the effort.
Prove the idea first?
PoC and R&D (Yes)
When the project or product calls for it, the proposal can proceed as a PoC for the client, or even an internal one, if there is no direct experience of the process yet. I then do R&D with the team: I work out what the technologies make possible, the type of delivery (cloud or on-premises), the requirements needed, the estimates and the costs.
Project (No)
When the proposal goes ahead as a project, I move on to management and organisation: kick-off, requirements, tender specification and deliveries.
+6sectors. Among them:Healthcare · Industry · Banking · Legal · Social · Design
Perspectives
AI without process is just a new cost.
Many companies are focused on implementing AI and pay little attention to understanding the process they want to improve. Without a diagnosis, AI adds costs instead of value.
The value is in the diagnosis, not the implementation.
Understanding where value is being lost, and whether AI is really the answer, is worth more than the technology you choose.
Before automating a process, it is worth asking whether it should exist at all.
AI / GenAI
I decide with data, not fashion. I define criteria and evals to choose models and libraries, and I stay hands-on: I build to validate ideas with clients.
Efficiency and speed, with a critical eye
With well-organised AI agents and a minimum of critical eye, you can get very far. At the ideation stage, that is enough: what counts is efficiency and speed, with critical thinking and strategic vision.
- Organise: I give the agents context, a goal and limits, so they work in the right direction.
- Question: I look at what comes out with a critical eye, before accepting it.
- Decide: The strategic vision is mine: what it is for, what it is worth and what to do next.
Tools change quickly. The responsibility to organise, question and decide stays.
Evaluate and build
Here too, these are the practices I use today, adjusted to each project.
Evaluate: Decide with data
Before proposing, I work out what is possible and what it costs. When there is no direct experience of the process yet, I do R&D with the team.
- Understand what possibilities the technologies offer
- Decide the type of delivery: cloud or on-premises
- Gather the necessary requirements
- Estimate effort and costs
- Define criteria and evals to choose models and libraries
- Take available infrastructure and compliance (GDPR) into account
Build: Show it working
I stay hands-on. Once the PoC has been awarded, I may build it myself or guide the team through it.
- Regular deliveries, to speed up the process
- Choose the best strategies with the team
- Show first (diagram, mockup) when it helps the client see the idea
- Validate each delivery with the client
- At the end, decide with the client whether to move on to a project
Before building, I show.
The client decides better when they see the idea. So I illustrate and validate with them before any investment: a drawing, a mockup, a PoC.
1. Process diagram
I draw how the work is done today (BPMN) and confirm with the client that I understood it correctly.
The client sees: The process, step by step. What is validated: That I understood how the work is done today.
2. Mockup
Wireframes and mockups of the screen, so the client can see and comment on the idea before there is any code.
The client sees: The solution’s screen. What is validated: That the idea makes sense to the people who will use it.
3. PoC
A small working version, with the client’s data, to test the idea before investing.
The client sees: The idea working. What is validated: That it works and is worth the investment.
Problems I have already solved this way
Scattered knowledge, slow answers
Virtual assistants that answer from the organisation’s own information, instead of making people go and look for it.
Hand-made proposals, always alike
Automatic proposal generation from tender specifications and technical product catalogues.
Finding similar opportunities
Identifying similar deals through visual recognition of products.
All of them started as PoCs: a way of validating the idea with the client before investing. Several became sold projects and solutions in use.
An AI core for clinical data in European Portuguese
Context
[HEALTH_PRODUCT] was born with a multilingual vision, starting in Portugal, with the aim of reducing the time healthcare professionals spend on clinical records.
Constraint
Sensitive clinical data, subject to GDPR and sector regulation, and inference on local infrastructure. Added to this is the scarcity of European Portuguese data that can be used responsibly and with appropriate licences, in a market where most models and datasets are based on Brazilian Portuguese.
Decision
- A modular pipeline of ASR, LLMs/SLMs and NER, with pseudonymisation of sensitive data before any reuse.
- Each model choice validated with purpose-built evals in European Portuguese, rather than generic benchmarks.
- An architecture ready for new languages without starting from scratch.
Result
By the users’ own estimates, a significant reduction in the time spent on note-taking. A pseudonymised database, ready for the future training of models.
My role
Led the product and the AI core for about a year, with a team of 6 people, working with stakeholders and clinical teams.
Several PoCs converted into sold projects.
The tools that support me: PyTorch · Hugging Face Transformers · sentence-transformers · spaCy · TorchAudio · vLLM · TensorRT · LlamaIndex · OpenAI API · Azure AI · Python · FastAPI · TypeScript · Node.js · React · React Native · SQL · PostgreSQL · Docker · Git · Claude · Claude Code · OpenCode.
Project Management
I run two to three consulting projects at once. With backlogs and roadmaps that balance business value and effort. Every delivery goes through functional testing before it reaches the client.
Scrum and Agile, applied and adapted
I know the Scrum and Agile models and apply them in an adapted way: not as a rigid standard, but shaped to each project, the team and the moment the business is in.
Starting point: A common starting point: tasks live in Excel spreadsheets. It is simple and works at first, but it limits visibility, traceability and pace as projects multiply.Evolved process: I have experience leading this evolution: moving task management into ClickUp and bringing in good Scrum and Agile practices, adjusted to each project, to speed up three fronts at once.
- Development: Sprints, a prioritised backlog and clearer deliveries.
- Management: Visibility of where each project stands, priorities and team capacity.
- Compliance: Traceability of tasks, decisions and owners.
Concepts I keep, even when adapted
- Sprints: Short work cycles, with a goal and deliverables.
- Dailys: Quick check-ins with the team.
- Reviews: Showing what has been done and gathering feedback.
- Retrospectives: Looking at the way of working and improving it.
Adapting does not mean abandoning the model: the purpose of each ceremony stays, the format adjusts to the project, the team and the moment the business is in.
From kick-off to go-live
Once approved, the path depends on whether it is a PoC or a project. These are the practices I use today. The format, order and pace are adjusted to each project.
If it is a PoC
I can build it myself or guide the team through building it, with regular deliveries, to speed up the process and choose the best strategies.
If it is a project
Kick-off
Starting meeting, covering the deliverables, the responsibilities and the associated risks.
Requirements workshops
Working sessions to define the requirements clearly.
Tender specification
Drawing up the tender specification.
Infrastructure and way of working
Defining the infrastructure and the way of working: what the sessions will look like and which methods to follow in developing the project.
Follow-up and deliveries
Follow-up sessions and deliveries every week, every fortnight or at another cadence, depending on the format and type of project. Alongside the client, I follow and speed up the process through to the testing phase.
Testing
When I have the availability, I run manual tests and help the team meet the requirements and improve the system.
Go-live
Final deliveries and going into production.
Adoption
The project is only finished when the solution is used. This is where training comes in.
Methods: Scrum, Kanban, backlog and roadmap management, team coordination.
Training & Adoption
A solution is only worth something if it is used. I run AI assessments and training in companies, with diagnostics, adoption roadmaps and prioritised projects.
From diagnosis to adoption
Assessment: Start with the process
Every assessment starts with the process, not the tool: identify where value is being lost and only then propose where AI helps.
The result: Diagnostic report; Adoption roadmap; Prioritised project list.
Training: Empower people
Training in AI fundamentals and in tools such as Claude, ChatGPT and Microsoft 365 Copilot.
Credential: Certified trainer (CCP).
Adoption: Make sure it is used
After go-live, training and follow-up help ensure the solution is actually used.
The aim: That the solution is actually used.
+30 assessments and training sessions, from webinars with 200 attendees to diagnostics in SMEs.
Skills map
Business Analysis
Requirements gathering · Stakeholder management · User stories and acceptance criteria · Use cases · BPMN · Wireframes · Functional specifications · Functional testing · Design Thinking · Requirements workshops
Project Management and Delivery
Pre-sales and proposals · Scoping and effort estimation · Planning · Prioritisation · Backlog and roadmap management · Team coordination · Agile · Scrum · Kanban · Tender specification · Risk management
Applied and Generative AI
Generative AI PoCs · AI agents · MCP · LLMs/SLMs · RAG · Embeddings · Virtual assistants · NLP · ASR · NER · Evals and model selection · On-premises inference · AI text generation · Visual recognition · European Portuguese
Strategy, Adoption and Compliance
AI assessments · Use case identification and prioritisation · Adoption roadmaps · AI training (CCP) · GDPR and health data · PII detection and pseudonymisation
AI Frameworks
PyTorch · Hugging Face Transformers · sentence-transformers · spaCy · TorchAudio · vLLM · TensorRT · LlamaIndex · OpenAI API · Azure AI
Development
Python · FastAPI · TypeScript · Node.js · React · React Native · SQL · PostgreSQL · Docker · Git
Tools
Claude · Claude Code · OpenCode · ChatGPT · Microsoft 365 Copilot · ClickUp · Notion · Figma · draw.io
Journey
Jan 2026 · CCP, Certificate of Pedagogical Competence
Certified trainer, IEFP.
Sep 2025 – Jul 2026 · Executive Master in Tech Leadership and Management
Porto Business School.
since May 2024 · Visual Thinking
AI Business Analyst & Project Manager since Jun 2025, after an internship in development and functional analysis.
Feb – Apr 2024 · Microprocessador
Software Developer Trainee: real-time data integration for industry, in direct contact with the client.
Sep 2022 – 2025 · Bachelor’s degree in Computer Engineering
ISPGAYA.
Jan 2018 – Jan 2024 · Groz-Beckert
Six years in industry; from 2022 alongside the degree.
Nov 2016 – Dec 2017 · ABB, PALJOR and FICO CABLES
An internship at ABB and then a varied period at other industrial companies, such as PALJOR and FICO CABLES.
2015 – Jan 2017 · ATEC
Specialist Technician in Automation, Robotics and Industrial Control.
Now: Looking for a new opportunity and studying: Hugging Face · Agents Course; Professional Scrum Master I · in preparation.
Languages: Portuguese (native) · English (advanced) · Spanish (basic)
Connect your agent
What is on this site is also available to your AI agent, through MCP. You can ask it to assess my profile, compare it with a job description or answer the questions of a first conversation.
Server address: https://portfolio-mcp.fabio-rafael7.workers.dev/
claude mcp add --transport http portfolio https://portfolio-mcp.fabio-rafael7.workers.dev/
Shall we talk?
LinkedIn is the best place to find me.