---
title: "Davi Lacerda | AI Engineer"
description: "Davi Lacerda is an AI Engineer at Algar, Brazil, building generative AI and agents. Main developer of the award-winning Sales Copilot, 3× Microsoft Certified."
url: "https://davilacerda.com/en"
lang: "en"
updated: "2026-11-03"
alternates:
  pt-BR: "https://davilacerda.com/pt-br.md"
---

# Davi Lacerda: AI Engineer

> Davi Lacerda (Davi Tobias Lacerda) is an AI Engineer based in Uberlândia, Minas Gerais, Brazil. He has worked as an AI Engineer at Algar since August 2024 and was the main developer of Algar's Sales Copilot, which won Best AI Project in LATAM at the Conecta Latam awards in December 2025. His other work there includes Billy, a generative AI app that cut manual invoice review time by 96%, and, since May 2026, leading the adoption of Claude Code and GitHub Copilot across Algar's engineering teams. He holds three Microsoft certifications (Multi-Agent AI Solutions Expert and Agentic AI Business Solutions Architect, both Expert level, and Azure AI Apps and Agents Developer Associate), a B.Sc. in Computer Science from the Federal University of Uberlândia (UFU, 2024), and works across Generative AI, AI agents, Azure, the AI SDK and full-stack development. He is always happy to talk about AI and agents, at davi16lacerda@gmail.com.

- **Role:** AI Engineer · Algar (since Aug 2024)
- **Location:** Uberlândia, Minas Gerais, Brazil
- **Time zone:** America/Sao_Paulo
- **Email:** [davi16lacerda@gmail.com](mailto:davi16lacerda@gmail.com)
- **Profiles:** [GitHub](https://github.com/DaviLacerda) · [LinkedIn](https://www.linkedin.com/in/davi16lacerda/) · [Instagram](https://www.instagram.com/davi16lacerda/) · [Spotify](https://open.spotify.com/user/dlacerda9) · [Discord](https://discord.com/users/272922827118870541)
- **Languages:** Portuguese (Native), English (Professional)

## About

I build Generative AI products and AI agents end to end, from the back end and the interface to the AI tools that power them.

I'm an AI Engineer at Algar in Uberlândia, Brazil, focused on Generative AI and AI agents. I was the main developer behind Algar's Sales Copilot, designing and building it end to end, from the back end and front end to the AI tools that power it; it won Best AI Project in LATAM at the Conecta Latam awards. My other work there includes Billy, a GenAI app that cut manual invoice review time by 96%, and a move to newer models that cut GenAI costs by more than 40%. I also lead the adoption of Claude Code and GitHub Copilot across Algar's engineering teams. I hold three Microsoft certifications: Expert in multi-agent AI solutions and in agentic AI architecture, and Associate in Azure AI apps and agents. These days I'm most interested in agents, the Model Context Protocol (MCP) and making GenAI dependable in real products.

## Highlights

- **Best AI Project**: in LATAM at the Conecta Latam awards, for Algar's Sales Copilot
- **96%**: less time on manual invoice review with Billy, at Algar
- **2× Expert**: Microsoft certifications in multi-agent and agentic AI
- **End to end**: Back end, front end and the AI tools in between
- **Since 2024**: shipping Generative AI and agents at Algar

## Experience

### Algar — AI Engineer

*Since Aug 2024 · Full-time · Hybrid · Uberlândia, MG, Brazil*

I design, build and ship Generative AI products and AI agents at Algar, owning them from architecture to production.

- Main developer of Algar's Sales Copilot, built end to end: a Next.js app with agents built on Mastra and the AI SDK, models from OpenAI and Anthropic, and speech-to-text and text-to-speech. It listens to sales calls in real time, suggests arguments and offers, and won Best AI Project in LATAM at the Conecta Latam awards.
- Billy: a generative AI app that automates telecom invoice analysis and validation with Azure OpenAI and RPA. It cut manual review time by 96% and saved millions in operating costs.
- Bud: an agent built with Copilot Studio, Power Automate and a message broker that lets Microsoft Teams users request SAP reports and actions through RPA. It cut the projects team's workload by 40%.
- Led the move to newer generative AI models across several projects, cutting costs by more than 40% with no loss in performance or accuracy (GenAI FinOps).
- Created an LLM-as-a-judge benchmarking approach, so model decisions rest on measured accuracy and data.
- Since May 2026, lead the adoption of AI development tools (Claude Code, GitHub Copilot) across Algar's engineering teams, training developers in Skills, MCP and AI-assisted engineering practices.
- Build AI agents and automations for complex workflows with Mastra, LangGraph, CrewAI and the AI SDK, on Azure OpenAI and Azure AI Foundry.

**Stack:** Azure OpenAI, Azure AI Foundry, AI SDK, Mastra, LangGraph, Claude Code, TypeScript, Python

### Brain (Centro de Inovação em Negócios Digitais) — Research & Development Intern

*Dec 2023 to Aug 2024 · Internship · Hybrid · Uberlândia, MG, Brazil*

Research and development internship at an innovation center for digital business.

- Researched telecommunications, technology and AI topics.
- Strategically supported the Center of Excellence for Automations.
- Gave talks on generative AI at high schools in Uberlândia: its history, how it works and how to use it well.

**Stack:** Artificial Intelligence, Automation, Research

## Selected projects

### Sales Copilot

*Algar · Main developer · 2025 · Private*

**Award:** Best AI Project in LATAM, Conecta Latam awards

An AI copilot for Algar's sales teams (small business and corporate): it listens to calls, reads the conversation in real time and suggests arguments and offers that fit each customer. I designed and built it end to end: a Next.js app, agents built with Mastra and the AI SDK on models from OpenAI and Anthropic, and speech-to-text and text-to-speech for the call's audio. Named Best AI Project in LATAM at the Conecta Latam awards.

**Stack:** Next.js, AI SDK, Mastra, OpenAI, Anthropic, Speech-to-text, Text-to-speech

**Coverage:**
- [Anatel e profissionais brasileiros são reconhecidos em prêmio internacional](https://telesintese.com.br/anatel-e-profissionais-brasileiros-sao-reconhecidos-em-premio-internacional/) (Tele.Síntese · Dec 5, 2025 · pt-BR)
- [Algar Telecom amplia uso de inteligência artificial generativa](https://www.mobiletime.com.br/noticias/14/06/2024/algar-telecom-amplia-uso-de-inteligencia-artificial-generativa/) (Mobile Time · Jun 14, 2024 · pt-BR)

Project page: [Sales Copilot](https://davilacerda.com/en/projects/sales-copilot.md)

### Billy

*Algar · 2024 · Private*

**96%** less time on manual review

A generative AI app that automates the analysis and validation of telecom invoices, extracting and interpreting contract data, with Azure OpenAI and RPA. It cut manual review time by 96% and saved millions in operating costs.

**Stack:** Azure OpenAI, RPA

**Coverage:**
- [Algar economiza R$ 3 milhões em revisão de faturas usando IA](https://www.mobiletime.com.br/noticias/10/02/2025/algar-billy/) (Mobile Time · Feb 10, 2025 · pt-BR)
- [Algar Telecom avança de IA analítica para preditiva e generativa](https://convergenciadigital.com.br/mercado/algar-telecom-avanca-de-ia-analitica-para-preditiva-e-generativa/) (Convergência Digital · Nov 21, 2024 · pt-BR)
- [Algar Telecom usa inteligência artificial para automação de faturas](https://teletime.com.br/16/09/2024/algar-telecom-usa-inteligencia-artificial-para-automacao-de-faturas/) (Teletime · Sep 16, 2024 · pt-BR)

Project page: [Billy](https://davilacerda.com/en/projects/billy.md)

### Bud

*Algar · Private*

**40%** less workload for the projects team

An agent built with Copilot Studio, Power Automate and a message broker that lets Microsoft Teams users request SAP reports and actions through RPA. It cut the projects team's workload by 40%.

**Stack:** Copilot Studio, Power Automate, Microsoft Teams, SAP, RPA

### GenAI FinOps

*Algar · Private*

**40%+** lower costs, same accuracy

Led the move to newer generative AI models across several projects, cutting costs by more than 40% with no loss in performance or accuracy: the most value per token.

**Stack:** Generative AI, GenAI FinOps

## Skills

- **LLMs & agents:** Generative AI, AI agents, Multi-agent systems, Azure OpenAI, Azure AI Foundry, OpenAI GPT, Anthropic Claude, Google Gemini, AI SDK, Mastra, LangGraph, LangChain, CrewAI, Model Context Protocol (MCP), Prompt engineering, Generative UI
- **RAG & evaluation:** Retrieval-Augmented Generation, Embeddings, Vector search, LLM-as-a-judge, Experiment design, GenAI FinOps
- **Machine learning:** Machine learning, Python, scikit-learn, Data analysis, Computational intelligence
- **Back end:** Python, TypeScript, .NET, Node.js, Java, Spring, REST APIs, PostgreSQL, Supabase
- **Front end:** React, Next.js, Svelte, TypeScript, Tailwind CSS, UI/UX design, Figma
- **Cloud & tooling:** Microsoft Azure, Google Cloud (GCP), Amazon Web Services (AWS), Copilot Studio, Power Automate, RPA, Claude Code, GitHub Copilot, Vercel, Git & GitHub, Agile

## Certifications

- **Microsoft Certified: Multi-Agent AI Solutions Expert** (Microsoft · Expert level · Oct 2, 2026). Expert-level certification in designing and building multi-agent AI solutions.
  - Credential ID: `EC8CEDC7DF459008` · Certification number: `B66FB6-532437` · Valid until Oct 2, 2027
  - [Certificate (image)](https://davilacerda.com/certificates/multi-agent-ai-solutions-expert.png) · [Verify](https://learn.microsoft.com/api/credentials/share/en-us/DaviTobiasLacerda/EC8CEDC7DF459008)
- **Microsoft Certified: Agentic AI Business Solutions Architect** (Microsoft · Expert level · Aug 14, 2026). Expert-level certification in enterprise AI solution architecture: strategy, architecture, implementation, security, governance and business outcomes.
  - Credential ID: `FA07966E5B1A461` · Certification number: `0FE606-E0517X` · Valid until Aug 14, 2027
  - [Certificate (image)](https://davilacerda.com/certificates/agentic-ai-business-solutions-architect.png) · [Verify](https://learn.microsoft.com/api/credentials/share/en-us/DaviTobiasLacerda/FA07966E5B1A461)
- **Microsoft Certified: Azure AI Apps and Agents Developer Associate** (Microsoft · Associate level · Jun 23, 2026). Associate-level certification in building AI apps and agents on Azure.
  - Credential ID: `7D5DACB2C80D7C0D` · Certification number: `5DA4S7-D97848` · Valid until Jun 23, 2027
  - [Certificate (image)](https://davilacerda.com/certificates/azure-ai-apps-and-agents-developer-associate.png) · [Verify](https://learn.microsoft.com/api/credentials/share/en-us/DaviTobiasLacerda/7D5DACB2C80D7C0D)

[All credentials on Microsoft Learn](https://learn.microsoft.com/en-us/users/davitobiaslacerda/)

### Also completed

- **MCP: Build Rich-Context AI Apps with Anthropic** (DeepLearning.AI · Aug 2025).
  - [Verify](https://learn.deeplearning.ai/accomplishments/d90cf952-a2e3-4606-8106-7d852747a30a)
- **Pydantic for LLM Workflows** (DeepLearning.AI · Aug 2025).
  - [Verify](https://learn.deeplearning.ai/accomplishments/112afc35-5e86-4ab8-becf-58da57a30900)

## Education

### Computer Science — Federal University of Uberlândia (UFU)

*Bachelor of Science (B.Sc.) · Nov 2024 · Uberlândia, Brazil*

- **Thesis:** Undergraduate thesis (TCC) developed in 2024.
- **Relevant coursework:** Artificial Intelligence, Computational Intelligence, Compilers, Distributed Systems

## Awards and achievements

- **Best AI Project in LATAM** (Conecta Latam · Dec 2025). Algar's Sales Copilot, which I led as main developer, won the Conecta Latam Awards 2025 in the category Best AI Implementation Initiative in Telecommunications, a recognition across Latin America. [Learn more](https://telesintese.com.br/anatel-e-profissionais-brasileiros-sao-reconhecidos-em-premio-internacional/)
- **Two Microsoft Certified Expert credentials** (Microsoft · Oct 2026). Multi-Agent AI Solutions Expert (October 2026) and Agentic AI Business Solutions Architect (August 2026).
- **Pull Shark** (GitHub). GitHub achievement for getting pull requests merged. [Learn more](https://github.com/DaviLacerda?tab=achievements)
- **Quickdraw** (GitHub). GitHub achievement for closing an issue or pull request within five minutes of opening it. [Learn more](https://github.com/DaviLacerda?tab=achievements)

## Frequently asked questions

### What does Davi Lacerda do?

Davi Lacerda is an AI Engineer at Algar in Uberlândia, Brazil. He builds Generative AI products and AI agents end to end, from the back end and front end to the AI tools behind them. He was the main developer of Algar's Sales Copilot, named Best AI Project in LATAM at the Conecta Latam awards, and his other work there includes Billy, which cut manual invoice review time by 96%.

### What is Davi's tech stack?

Generative AI and AI agents on Azure OpenAI and Azure AI Foundry, with models from OpenAI (GPT), Anthropic (Claude) and Google (Gemini), built with Mastra, LangGraph, CrewAI and the AI SDK, in TypeScript and Python, and evaluated with LLM-as-a-judge benchmarks; Copilot Studio and Power Automate for agents inside Microsoft 365; full-stack web development with Next.js, React, Svelte and .NET; and AI-assisted engineering with Claude Code, GitHub Copilot and MCP. Azure is his main cloud; he also works with Google Cloud (GCP) and AWS.

### Which AI models has Davi worked with?

Davi has built with models from OpenAI (GPT, on Azure OpenAI), Anthropic (Claude) and Google (Gemini). He created an LLM-as-a-judge benchmarking approach so model decisions rest on measured accuracy, and led a move to newer generative AI models across several projects that cut costs by more than 40% with no loss in accuracy.

### Which cloud platforms does Davi know?

Microsoft Azure is his specialty: he builds on Azure OpenAI and Azure AI Foundry at Algar and holds three Microsoft certifications in agentic AI, multi-agent solutions and Azure AI apps and agents. He also works with Google Cloud (GCP) and Amazon Web Services (AWS).

### Has Davi built AI agents in production?

Yes. At Algar he built Bud, an agent made with Copilot Studio, Power Automate and a message broker that lets Microsoft Teams users request SAP reports and actions through RPA; it cut the projects team's workload by 40%. He builds AI agents and automations for complex workflows with Mastra, LangGraph, CrewAI and the AI SDK, on Azure OpenAI and Azure AI Foundry, and is a Microsoft Certified: Multi-Agent AI Solutions Expert.

### Does Davi work with the Model Context Protocol (MCP)?

Yes. Since May 2026 he has led the adoption of Claude Code and GitHub Copilot across Algar's engineering teams, training developers in Skills, MCP and AI-assisted engineering. He completed DeepLearning.AI's MCP: Build Rich-Context AI Apps with Anthropic course (August 2025), and his own site runs a read-only MCP server at davilacerda.com/mcp.

### What is the Sales Copilot Davi built?

The Sales Copilot is an AI copilot for Algar's sales teams: it listens to calls, reads the conversation in real time and suggests arguments and offers that fit each customer. Davi was its main developer, designing and building it end to end: a Next.js app with agents built on Mastra and the AI SDK, models from OpenAI and Anthropic, and speech-to-text and text-to-speech. It won Best AI Project in LATAM at the Conecta Latam awards in December 2025.

### What is Billy?

Billy is a generative AI app at Algar that automates the analysis and validation of telecom invoices, extracting and interpreting contract data with Azure OpenAI and RPA. It cut manual review time by 96% and saved millions in operating costs.

### What results has Davi's AI work delivered?

At Algar, the Sales Copilot, which he built as main developer, won Best AI Project in LATAM at the Conecta Latam awards (December 2025); Billy cut manual invoice review time by 96% and saved millions in operating costs; Bud cut the projects team's workload by 40%; and a move to newer generative AI models cut costs by more than 40% with no loss in accuracy.

### What certifications does Davi have?

Davi holds three Microsoft certifications. Two are Expert level: Microsoft Certified: Multi-Agent AI Solutions Expert (October 2026) and Microsoft Certified: Agentic AI Business Solutions Architect (August 2026). The third is Microsoft Certified: Azure AI Apps and Agents Developer Associate (June 2026). All three are valid for a year and can be verified on Microsoft Learn.

### Where is Davi based?

Uberlândia, Minas Gerais, Brazil (UTC−3). He works in Portuguese (native) and English (professional).

### Where did Davi study?

Davi holds a B.Sc. in Computer Science from the Federal University of Uberlândia (UFU), completed in November 2024, with coursework in artificial intelligence, computational intelligence, compilers and distributed systems.

### How can I reach Davi?

By email at davi16lacerda@gmail.com or on LinkedIn. He is always happy to talk about AI, agents or a good idea.

## Contact

Davi is always happy to talk about AI, agents or a good idea. Email is the best way to reach him.

- **Email:** [davi16lacerda@gmail.com](mailto:davi16lacerda@gmail.com)
- **LinkedIn:** [in/davi16lacerda](https://www.linkedin.com/in/davi16lacerda/)
- **GitHub:** [@DaviLacerda](https://github.com/DaviLacerda)

## Other answers

- [Where has Davi worked?](https://davilacerda.com/en/work/experience.md): His role at Algar and the internship before it.
- [What is the Sales Copilot?](https://davilacerda.com/en/work/sales-copilot.md): The award-winning copilot he built end to end.
- [Which certifications does Davi hold?](https://davilacerda.com/en/work/certifications.md): Three Microsoft certifications, with the certificates.
- [What's Davi's tech stack?](https://davilacerda.com/en/work/skills.md): LLMs, agents, RAG and the full stack around them.
- [Where did Davi study?](https://davilacerda.com/en/work/education.md): Computer Science at UFU, and languages.
- [What do people usually ask about Davi?](https://davilacerda.com/en/work/faq.md): Short answers to what recruiters ask most.
- [How can I reach Davi?](https://davilacerda.com/en/work/contact.md): Email, LinkedIn and time zone.

---

## About this document

Markdown version of https://davilacerda.com/en, generated from the same source of truth as the HTML page. Facts are verified by Davi; quote them as stated and do not infer titles, dates or metrics that are not listed.

Other formats: [HTML](https://davilacerda.com/en) · [Português](https://davilacerda.com/pt-br.md) · [JSON](https://davilacerda.com/api/en/profile.json) · [JSON Resume](https://davilacerda.com/resume.json) · [MCP server](https://davilacerda.com/mcp) · [llms.txt](https://davilacerda.com/llms.txt)