Derick Bessa – Resume

Professional Summary

Full-Stack Developer with experience in React, Next.js, Angular, and React Native on the frontend; Node.js, Java/Spring Boot, and Python/FastAPI on the backend. Applied research background in computer vision, integrating AI models into end-to-end mobile and web systems. Founder of TáNaLista, an event management SaaS with QR Code check-in.

Education & Languages

B.Sc. in Computer Science – IFCE (Federal Institute of Ceará), Fortaleza 2024 – 2028

Languages: English (Advanced) · Portuguese (Native)

Technical Skills

Frontend: React.js, Next.js, Angular, React Native, TypeScript, Tailwind CSS

Backend: Node.js, Java/Spring Boot, Python/FastAPI, REST APIs, JWT/OAuth2

Databases: PostgreSQL, SQLite, MySQL

AI & Computer Vision: YOLOv8/v11, OpenCV, MediaPipe, PaddleOCR, scikit-learn, pandas

DevOps & Tools: Docker, Git, GitHub Actions, Vercel, Swagger/OpenAPI

Professional Experience

LabVicia – IFCE, Fortaleza

Sep 2025 – Present

Full Stack Developer · Research Scholar

Implemented server-side pagination and eliminated redundant API calls, reducing data transfer volume and making table loading noticeably faster for end users.
Refactored the inbox module with infinite scroll and automatic message updates, eliminating manual refreshes and making the communication flow continuous and uninterrupted.
Delivered full internationalization for 4 languages (PT, EN, ES, DE) with dynamic switching via ngx-translate, enabling international teams to use the system with no rebuild or redeployment required.
Standardized components with Angular Material across the entire system, reducing visual inconsistencies and speeding up new screen development for the team.

LAPISCO – Research Lab at IFCE, Fortaleza

Jun 2024 – Mar 2026

Research Scholar · Computer Vision & Full Stack Development

Integrated detection and classification models (YOLOv8/v11, MediaPipe, OpenCV) with the Java/Spring Boot backend via a Python/FastAPI layer, enabling real-time processing of frames captured by the camera in the React Native app.
Built dashboards and technical reports using pandas and scikit-learn to monitor model training, allowing the team to identify bottlenecks and tune hyperparameters based on concrete data.
Refactored Java backend APIs with a focus on separation of concerns and object-oriented design, improving maintainability and reducing onboarding time for new team members.
Worked across the full AI pipeline — from image capture in the React Native frontend to result consolidation in the Java backend — ensuring traceability and data consistency across all layers.

Gomes & Araújo Law Firm – Freelance

2025

Front-End Developer

Delivered a complete landing page in React 19 + Vite + Tailwind CSS with 8 sections and a fully responsive layout, with no external UI library dependencies, resulting in a lighter bundle and faster load times.
Built a Node.js/Express backend with transactional email delivery (Resend API) and a WhatsApp CTA, converting page visitors into leads directly without any external CRM tool.

Projects

SFA – Automotive Inspection System (License Plate Reader)

Automatic vehicle license plate recognition

Achieved 97% accuracy in license plate recognition by combining detection via YOLOv8 (custom-trained model) with OpenCV preprocessing and PaddleOCR reading, making the system reliable for real-world enforcement use.
Implemented support for three plate formats (old standard, Mercosul, and motorcycle) with per-format character correction logic and automatic image rotation, eliminating failures caused by incorrect camera angles.
Exposed the pipeline as a Python/FastAPI layer integrated with the Java/Spring Boot backend, allowing computer vision processing to run in a decoupled manner with no impact on the main system's performance.

Libras Translator

Brazilian Sign Language recognition from static images

Built a complete ML pipeline in Python — collecting a custom dataset with 26 classes, extracting hand landmarks via MediaPipe Hands, and training a Random Forest classifier with scikit-learn — delivering a functional model for recognizing the Libras alphabet.
Applied relative coordinate normalization to ensure scale and position invariance, making the model robust to variations in camera distance and framing without the need for retraining.
Implemented an inference module with OpenCV to visualize landmarks and overlay predictions on the image, enabling fast visual model validation without a dedicated interface.

TáNaLista

Event management SaaS with QR Code check-in

Built a SaaS platform in Node.js, Express, and PostgreSQL with Prisma ORM for QR Code check-in and real-time guest tracking, eliminating manual lists and reducing event entry time for organizers.
Implemented JWT authentication with bcrypt and a modular architecture, ensuring secure access control and enabling each system module to evolve independently.
Configured CI/CD with GitHub Actions and automated deployment to Vercel, eliminating manual deploys and ensuring every merge to main is immediately live in production.