Level up your organization.

AI and data systems that deliver measurable results.

I'm Emanuel Elias, an AI and data consultant. I help organizations find where intelligence pays off, then lead the work from diagnosis and architecture to a system running in production, with the security and governance your business requires. I stay accountable for the result. On site in Belo Horizonte and the surrounding region, or remote worldwide.

Book a call See selected work

  • 8+ years building applied AI and data systems
  • Postgraduate degree in Artificial Intelligence, Federal University of Viçosa
  • Computer Engineering degree, Federal Institute of Minas Gerais
  • Hands-on with Azure, AWS and GCP, and fully private on-premise environments

Where I usually step in

Decisions wait on data nobody trusts

Reports disagree, spreadsheets multiply and every answer takes days. I build the data foundation and the dashboards leaders actually use.

Manual work is eating margin

People still collect, reconcile and report by hand. I automate it end to end, including data from sites and systems that have no API.

AI pilots never reach production

The demo works, then stalls on security, cost or integration. I take it through deployment, monitoring and handover.

Forecasts that can't be defended

Demand, price or risk predictions without a clear method. I build models you can explain, test and monitor.

Three ways to work together

Diagnostic

For leadership teams deciding where AI and data will pay off.

  • Digital maturity assessment of your data, systems and processes, with surveys and on-site visits
  • Opportunities ranked by impact and effort
  • Roadmap with architecture, risks and next steps

Typically 2 to 4 weeks

Build and deploy

For teams with a defined problem and a deadline.

  • Working proof of concept on your real data
  • Production system on your cloud or on premise
  • Tests, documentation and team handover

Typically 6 to 16 weeks

Advisory and operations

For companies that want a senior partner on call.

  • Monthly technical leadership and review
  • Monitoring, cost control and model upkeep
  • Priority access for new requests

Monthly retainer

Every engagement starts with a written scope and fixed deliverables. Pricing is proposed per project after the first conversation.

What I build

Predictive models

Forecasting, anomaly detection, classification and computer vision.

Generative AI and language

Assistants, search across internal documents, text and sentiment analysis, and evaluation of model quality.

Data platforms

Lakehouse design, ELT pipelines, data quality and governance.

Automation and data collection

Workflow and executive report automation, web data collection and integration between systems.

Analytics and dashboards

Descriptive, predictive and prescriptive reporting in Power BI, Tableau, Looker or Streamlit.

Cloud, on-premise and MLOps

Automated releases, containers and monitoring on Azure, AWS and Google Cloud, or on your own servers.

Tools I use most: Python, SQL, dbt, Spark, Databricks, Airflow, Docker, Kubernetes, Azure, AWS, Google Cloud, Power BI

Selected work

Client names are confidential. Details are shared under NDA.

Generative AI platform for a national institution

Challenge
Teams needed reliable answers from large internal knowledge sources, and leadership needed visibility into what the system was doing.
Approach
Built an LLM application end to end, with separate development, staging and production environments, automated releases and secure deployment.
Result
A governed assistant running in production, ready to be monitored and curated over time.

Observability and curation of AI agents for a national institution

Challenge
As AI agents spread across the institution, leadership needed visibility into how they were used, what they cost and how well they answered, plus clear rules to keep them under control.
Approach
Built an observability and curation layer with analysis of user feedback, access and usage metrics, macro and per-resource cost analysis, and the topics users search for most. Added warning triggers that fire when monthly spending reaches a set value, budget locks that apply when the pre-set limit is reached, and guardrails.
Result
One view to manage the agents: what people ask, how well the agents answer and what each resource costs, with alerts when monthly spending reaches the set value and budget locks when the pre-set limit is hit.

Budget and accounting dashboard for the Deliberative Council of a national institution

Challenge
The Deliberative Council needed to follow technology spending in budget and accounting terms, separating investment from operating cost.
Approach
Designed an executive dashboard that consolidates AI and data platform costs and presents them as CAPEX and OPEX for budget and accounting management.
Result
The council uses it every day, with a clear, decision-ready view of what is invested and what it costs to operate.

Strategic indicators, KPIs and OKRs: management and measurement

Challenge
Strategic indicators are often calculated with methodological errors and inconsistent data handling, which undermines trust in the numbers leadership relies on.
Approach
Managed the full cycle end to end: requirements gathering, data mart construction, methodological review of indicators that already existed, then measurement, calculation, validation and sign-off, and presentation in dashboards.
Result
Indicators leadership can trust: reviewed methodology, validated figures and dashboards built to support decisions.

Digital maturity assessment and data-driven roadmap

Challenge
The organization needed to understand how mature it was digitally and how to build a more data-driven culture.
Approach
Measured digital maturity through surveys, on-site visits and information gathering, produced a diagnostic and drew a roadmap for a more data-driven culture and digital transformation.
Result
A diagnostic and a roadmap, so leadership knows where the organization stands and what to do next.

Brand image and sentiment analysis

Challenge
What do other people think of your brand? The answer is spread across news portals, social networks and countless comments.
Approach
Analyzed the image of the organization in news portals and digital media: comments and reactions, main topics and sentiment analysis.
Result
A clear picture of public perception, showing which topics drive the conversation and how people feel about them.
Show more projects

Automation of executive reports

Challenge
Every month a company had to produce 29 executive reports, each with specific filters for every branch, which multiplied the total even further.
Approach
Automated the production of the reports and of the descriptive analysis of the figures.
Result
The technical team no longer spends its time on repetitive reporting and can focus on what matters: delivering results and adding value.

Lakehouse and automated data pipelines

Challenge
Data scattered across systems, extracted by hand, with no single trusted source for strategic indicators.
Approach
Designed a medallion architecture (raw, cleansed, business-ready) with automated ELT, quality checks and scheduled orchestration.
Result
One governed source of indicators feeding dashboards without manual rework.

Market intelligence and forecasting

Challenge
Price movements are influenced by news and public sentiment that traditional models ignore.
Approach
Automated collection of news and social data, Portuguese-language sentiment analysis, and time series and LSTM models combining market data with sentiment.
Result
A repeatable forecasting pipeline that measures how much sentiment adds to prediction quality.

Computer vision for medical imaging

Challenge
Reading images by eye is slow and varies between specialists. Research projects.
Approach
Built a C++ and OpenCV tool that classifies and measures anomalies in MRI scans, and a CNN for malaria detection in blood smear images with improved preprocessing.
Result
Consistent, automated measurements designed to run efficiently without heavy frameworks.

Also: industrial automation on low-cost hardware, time series decomposition, public policy data analysis.

How an engagement runs

  1. Discover

    An initial 30-minute conversation about your goal, data and constraints. You get a written summary and an honest recommendation, even when the answer is not to hire me.

  2. Prove

    A small, time-boxed test on your real data to confirm the idea works before you commit a larger budget.

  3. Build

    Iterative delivery with regular demos, on your infrastructure, with security and cost in view from day one.

  4. Hand over

    Documentation, training and monitoring so your team can run it. Ongoing support is optional.

Project leadership for executive sponsors

Beyond the technical work, I take responsibility for the outcome: aligning the business case, running the project and keeping decision makers informed.

Business case and scope

Feasibility analysis, requirements gathering and a clear view of impact, effort and risk before any budget is committed.

Planning and governance

Scope, schedule, risks and dependencies tracked in the open, with decisions and changes recorded for the steering committee.

Executive communication

Concise status reports and data storytelling that turn technical progress into decisions your leadership can take.

Team and vendor coordination

I lead technical teams and align business areas, IT, security and other vendors so the project keeps moving.

About Emanuel

I'm a computer engineer with a postgraduate degree in artificial intelligence. Since 2018 I have led data and AI work for organizations that cannot afford to get it wrong, from institutions with national reach to international engagements. My projects cover the full path: diagnosis, architecture, forecasting models, automation, executive dashboards and delivery.

I work as a senior advisor to executives and technical teams alike. I turn business questions into measurable results, with governance, security and compliance built in from day one, including fully private, on-premise environments when confidentiality demands it.

Based in Belo Horizonte, I serve Belo Horizonte and the surrounding region on site, and clients anywhere in the world remotely.

Education

  • Postgraduate degree in Artificial Intelligence and Computational Intelligence, Federal University of Viçosa
  • Bachelor's degree in Computer Engineering, Federal Institute of Minas Gerais, Bambuí campus

How I work

On site or remote

In Belo Horizonte and the surrounding region, or remotely with teams anywhere, using video calls, shared repositories and your project tools.

Confidential by default

NDA before any data is shared. Work happens inside your environment whenever possible.

Compliance by design

Every solution is designed to meet the data protection and security obligations that apply to your organization, such as GDPR and LGPD.

Senior and hands on

You work directly with me, not with a rotating team.

Questions

Do you work remotely?

Yes. Most projects run fully remote. I also work on site in Belo Horizonte and nearby, and elsewhere when it adds value, with travel agreed in advance.

How do you handle confidentiality and data security?

I sign an NDA before seeing any data, work inside your environment with least-privilege access, and align designs with the regulations that apply to you, such as GDPR and LGPD. No client data is sent to third-party models without your written approval. For the most sensitive cases, I build custom on-premise solutions.

What kind of companies do you work with?

Mid-size and large companies, institutions and well-funded startups. I prefer projects with a clear business owner and a measurable goal.

Do we need to have our data ready?

No. The diagnostic exists to find out what you have, what is usable and what it would take to get value from it.

How is pricing structured?

Fixed-scope proposals for diagnostics and builds, and a monthly retainer for advisory. You receive a written proposal after the first conversation.

Which time zones can you cover?

I'm in Brazil (UTC−3), with comfortable overlap for the Americas and Europe. For Asia-Pacific, calls are scheduled early in the morning or in the evening.

Can you also manage the project, not only build it?

Yes. I can act as technical lead and project manager, owning scope, schedule, risks and budget tracking, with regular reports for sponsors. I also coordinate your team and other vendors.

Tell me what you want to improve.

Share a few lines about your goal. You will get a reply within one business day.

Prefer to write directly?

What do you need?