Douglas Trajano

A Python library that introduces Agent Skills into the Pydantic AI framework.

pydantic-ai-skills is a Python library that introduces Agent Skills into the Pydantic AI framework.

• 260+ GitHub stars (released December 2025).
• 260k+ monthly downloads on PyPI.

A toxicity detection system based on OLID-BR.

ToChiquinho is a toxicity detection system based on OLID-BR, a dataset for Offensive Language Identification in Brazilian Portuguese.

The system is composed of 5 models, each one specialized in a different task in hate speech domain.

Data App with advanced analytics to improve the quality of chatbots.

Anallyticabot is a platform that provides advanced analytics to help you improve your chatbot.

Identify and improve your virtual assistant’s experiences easily and quickly.

Anallyticabot is free to use and is an open-source project!

Technology

Leading the design, development, code review, and technical mentorship of AI/ML solutions for the OneAI team at ADP, collaborating with product and engineering teams to drive strategic initiatives across ADP’s key products, including DataCloud, Workforce Now, and Lyric.

• I develop ADP Assist for Analytics, a conversational experience within ADP’s products that leverages GenAI to provide users with insightful and accurate analytics based on their HR data. It was recognized with the Generative AI Innovation Award in the 2024 AI Breakthrough Awards.
• Developed a serverless search engine for ADP Reporting combining Amazon OpenSearch Service and Language Models to perform hybrid search and enhance the search capabilities of fields and filters when creating reports.

Financial Services

Work on all aspects of the design, development, and delivery of data science and analytics solutions to solve complex business problems and drive product strategies. Provide technical leadership and mentorship to less experienced team members.

• Developed a Text Clustering System that processes 2,3M messages monthly, accelerating the chatbot development by 30% by showing the most relevant unanswered topics.
• Authored an Automatic Speech Recognition (ASR) pipeline, powered by OpenAI Whisper Large V3, that transcribes up to 70k customer service calls monthly enriched with emotion recognition, speaker identification, and automated quality monitoring.
• Developed GenAI applications powered by LangChain and Databricks MLflow to generate synthetic datasets, make classifications using Zero and Few-Shot Learning and summarize text data.
• Developed a Multi-Class Text Classifier that identifies users trying to force a human transfer and helps the chatbot act accordingly, improving the retention rate by 2.24%.

Technology

Led a data team of 10+ data engineers and data analysts, responsible for the data of OMEN Gaming Hub, a software that provides a unified gaming experience for HP customers. Directly responsible for the development of machine learning models used by HP gaming apps.

• Supported a toxicity detection system to ensure health usernames in OMEN Gaming Hub, with approximately 50k predictions per month.
• Redesigned data pipelines to support near real-time ingestion, introduced data quality checks and reduced compute costs by +90%.
• Developed a hybrid DS/ML platform for on-premises and cloud clusters powered by Kubernetes and Istio, leveraging tools such as JupyterHub, Apache Spark, MLflow, Boxkite, MinIO, Grafana, Redash, etc.
• Led external research projects in partnership with universities and research institutes such as Purdue University, SENAI CIMATEC, Instituto Atlantico, and others.

Financial Services

Development of data science solutions for customer service in conversational channels such as WhatsApp and WebChat, working throughout the entire life cycle (design, development, deployment, and support).

Highlights:

• Developed an Intent Recommendation NLP Pipeline that filters relevant messages and clusters them into new intents, enabling developers to create new intents without the need to manually label the data.
• Developed a Learning to Rank model to suggest relevant intents for the welcome message of the virtual assistant based on the customer’s profile.
• Worked with other data scientists to develop a methodology and tools to democratize data science inside Sicredi. Specifically, I wrote a Python library to better integrate our models with the MLOps tools used by the company.
• Developed a web application that generates advanced reports about the performance of the chatbot’s models, using techniques like cross-validation and integrations with MongoDB and APIs, etc.

Technology

Project-based work to develop and support machine learning enabled solutions for companies in different markets and contexts.

• Developed an NLP Pipeline that classifies legal actions into predefined categories, and adds them to the customer CRM (Salesforce) assigning each case (action) to the respective company area based on model predictions.

Technology

Chatbot developer contracted to Sicredi project. Development of intents, entities, and conversational flows in IBM Watson Assistant.

Highlights:

• Developed conversational flows for financial products like credit, account bank, credit recovery, etc. Including integration with internal systems to provide transactional information to the customer.
• Developed Python Scripts for data processing and data analysis. Developed scripts to evaluate the metrics of the chatbot and suggest improvements to reduce the number of conflicts in the chatbot intents.
• Advocated for the use of artificial intelligence in the company, providing technical training and explaining the benefits and disadvantages of the technology.
• Helped to define the content process used by the team.

Technology

As a Tech Support Specialist, I was responsible for managing service queues and workflows using Sprinklr, a Social CRM tool, for social media teams distributed in Latin America (Brazil, Costa Rica, and Panama). I also developed dashboards and reports for executives and managers, keeping them informed about the performance of the teams.

Highlights:

• Provided technical support for Dell products on social media (Facebook, Twitter, etc.).
• Created and maintained knowledge base articles for troubleshooting issues and providing solutions.
• Developed Analytics Reports and Dashboards for executives and managers for better decision-making.
• Developed Dell’s chatbot (Tell) for social media teams in Brazil and Costa Rica.
• Created automated workflows in Sprinklr to distribute and escalate incidents across global teams.

As a Tech Support Associate, I provide technical support and customer care for Dell products, working on executive escalations of complex issues and sensitive cases to Dell’s brand image. I also worked throughout the refund and exchange process for Dell products.

Highlights:

• Provided technical support via phone and email for Dell products (notebooks, desktops, tablets, etc.).
• Automated routine tasks in the tech support team (rewarded with a bronze award).
• Maintained the high standards of quality and productivity set by the Dell support team.
• Managed refund and exchange process for Dell products.

Focus on Artificial Intelligence