Technology consulting · Remote, worldwide
I'm Jonathan Bispo. For more than 15 years I have designed, built and shipped platforms that have to work for real: high data volumes, complex integrations, telephony, mobile, cloud and, today, AI applied to the business — agents, LLMs, automation and models trained for a specific domain.
years of software development
years leading teams as CTO and architect
years of AWS cloud architecture
product platforms kept running in production
Services
I work from the architecture design all the way to the code in production — on my own or alongside your team.
Designing systems from scratch or reviewing existing architectures: domain modelling, service boundaries, API contracts, database choices, queues and scaling strategy.
Putting AI inside your product in a useful and controlled way: LLM agents, RAG over your own data, MCP servers, process automation and voice with real-time transcription and synthesis.
Software built around how your operation actually works, not the other way around. Web platforms, mobile apps, back-offices, system integrations and internal automation.
Legacy system that is slow, expensive or hard to evolve? I run a technical assessment, prioritise what pays off quickly and lead the modernisation without stopping your operation.
Lean, predictable infrastructure on AWS: containers, CI/CD, observability, backups, VPN and cloud cost reduction.
From collection to model: data pipelines, analytics dashboards, automatic classification of open-ended text and machine learning models trained for your domain.
How I work
You understand every stage, what you get and what it costs before anything starts.
A 30 to 60 minute conversation to understand the business problem, the current technical setup and what has already been tried. Free of charge.
You get a document with the solution design, scope, risks, timeline and investment. In plain language, reviewable by people outside of tech.
I work in short cycles with demonstrable deliverables. You follow the progress and can adjust priorities at every cycle.
Going live is part of the job: deployment, monitoring, documentation and knowledge transfer to your team.
Projects
Real systems, with real users, running every single day.
Market research · SaaS
Full platform for questionnaire design, field collection through mobile and web apps, quota control, asynchronous processing of large volumes and analytical export. Django + PostgreSQL + AWS S3 architecture with background workers.
Voice AI · Telephony
Voice application on top of Asterisk (ARI + ExternalMedia/RTP) with a real-time pipeline of transcription, LLM natural language understanding and speech synthesis — running phone interviews with no human operator. Includes a predictive dialler and a supervision dashboard.
AI agents · Automation
A custom MCP server authenticated against the product's own database, exposing safe tools scoped to each user, plus agents that monitor quotas, trigger alerts and run operational routines. Integrated with WhatsApp and n8n automations.
Logistics · Traceability
A suite of modules for order management, picking, tracking and sanitisation reporting, with a desktop app for shop-floor operation, barcode scanning and report generation. Node.js/LoopBack service architecture on MongoDB, with versioned shared packages.
Telecom · Quality
Automated test tooling for CPEs (customer premises equipment) with bench execution, test scripts, batch and sample control, multi-level approval/rejection and quality dashboards — connected over VPN to remote labs.
Machine Learning · NLP
A BERT fine-tuning pipeline in PyTorch for multiclass classification of open-ended answers, preserving the context across questions from the same respondent — replacing manual coding with automatic classification.
Tech stack
I pick tools by the problem, not by the hype. These are the ones I run in production.
About me
I started teaching computing in 2005 and never stopped: I taught programming in technical courses and lectured on mobile development in PUC Minas' postgraduate IoT programme. In parallel I built a technical career up to Chief Technology Officer and Software Architect, a role I have held since 2016 at product companies.
I hold an MSc in Biomedical Engineering from UFMG, with research on embedded systems, instrumentation and neuromuscular diagnosis, and a BSc in Computer Engineering from PUC Minas. That background gave me something I use every day in consulting: method. Before writing code, understand the problem, measure, and decide with data.
I have a declared preference for functional programming and design my systems with those concepts in mind — predictable, testable code that is easy to evolve. And ever since LLMs became a production tool, I have been integrating AI into real platforms: agents, voice, automation and models trained for specific domains.
Frequently asked questions
It starts with an assessment of the current system or of the business problem. From there I deliver the solution design (components, data, integrations, infrastructure and risks) and I can lead the implementation, support your team, or both. The format adapts to the size of the project.
It means putting AI where it removes a real cost or bottleneck: an agent that answers and qualifies customers, intelligent search over your documents and databases (RAG), automatic text classification, voice support, or automation that replaces repetitive manual work. I assess where the return is real and implement it with cost and quality under control.
Yes. I work both with small companies that need their first custom system and with product companies that need architecture, modernisation or a senior specialist next to the team.
I work remotely with clients anywhere, and on site in Belo Horizonte and the metropolitan region of Minas Gerais, Brazil, whenever the project calls for it.
It depends on the format: a fixed-scope project, a monthly retainer of hours for ongoing consulting, or a one-off assessment delivered as a technical report. The proposal states exactly what is included before you decide.
Yes, that is a common scenario. I start with a technical audit of what exists (code, database, infrastructure and cost) and present the path: keep and evolve, refactor in parts, or rebuild. The recommendation comes with reasoning and numbers, not opinion.
Contact
Tell me in a few lines what you need. I answer personally.