Thiago Deeke Viek
São Paulo
Summary
As a self-taught software engineer with a strong background in AI, I specialize in developing cross-platform applications and enhancing them with Deep Learning models trained on real-world data.
Overview
8
8
years of professional experience
Work History
Lead Software Engineer
Inlevel GmbH
11.2022 - Current
- - Cross-Platform Development: Architected and implemented a robust frontend using Flutter, adhering to Clean Architecture principles and leveraging Bloc for efficient state management
- Integrated real-time data and user authentication through a Supabase backend. - Custom Microservices: Designed and deployed custom microservices utilizing AWS API Gateway, SNS, SQS, and Lambda functions
- Implemented a secure authorization process and automated infrastructure deployment with CloudFormation. - LLM Agent Orchestration: Orchestrated service pipelines with LLM Agents, employing advanced prompt engineering techniques for precise agent definition and task execution. - Model Fine-Tuning: Fine-tuned 8B LLaMa instruction models, optimizing them for task-specific performance within the application. - Scalable System Design: Engineered the system architecture to support high scalability. - Database Management: Managed PostgreSQL as the primary database for structured data and employed MongoDB as an auxiliary database for unstructured data
- Additionally, utilized Neo4j for sophisticated data presentation and graph-based analysis.
Software Engineer
Inlevel GmbH
07.2020 - 11.2022
- - Implemented and tested various computer vision model architectures for object detection, segmentation, clustering, and image classification, optimizing them for real-time performance - Worked with state-of-the-art deep learning architectures including CNNs, GANs, and Transformers using PyTorch, applying pruning and optimization techniques for efficient execution - Assisted in designing and deploying MLOps pipelines to improve model development and deployment processes, primarily for on-premise environments - Collaborated on creating inference acceleration engines using C++/CUDA and Docker for production environments, ensuring millisecond latency performance - Participated in R&D project management - Supported software development and product design efforts
Research Assistant
Werkzeugmaschinenlabor, WZL der RWTH Aachen
10.2019 - 07.2020
- - Implemented and evaluated multiple cutting-edge Convolutional Neural Network (ConvNet) architectures to accurately identify worn cutting tool edges - Achieved real-time segmentation results with an Intersection over Union (IoU) of 0.6 and mean Average Precision (mAP) of 0.25 using Mask R-CNN network - Conducted comparative analysis of various SOTA ConvNet models to determine the most effective architecture for the specific use case - Collaborated with cross-functional teams to integrate the computer vision system into existing manufacturing processes
Data Scientist
RUDOLPH
08.2018 - 02.2019
- - Developed time-series-based regression models for predictive maintenance - Achieved a final Mean Relative Error of 18% in model performance - Enhanced early detection capabilities for potential process faults - Collaborated with cross-functional teams to optimize model accuracy - Gained practical experience in applying data science to real-world manufacturing challenges
Project Manager
PET - MA
08.2017 - 12.2017
- As the Project Manager of the group, I monitored the group's performance regarding project progress, utilizing PMBOK and SCRUM methodologies
- Additionally, I developed a risk management software with MATLAB (Frontend and Backend) to evaluate execution time variance for project tasks, enhancing our project planning capabilities.
Research Assistant
Werkzeugmaschinenlabor, WZL der RWTH Aachen
08.2016 - 09.2017
- - Utilized MATLAB and Simulink to create a robust data processing pipeline - Implemented algorithms to merge simulation data from SPARTApro with live operational streams - Developed real-time tool wear assessment functionality using statistical analysis techniques - Optimized data flow between SPARTApro (specialized Gear Hobbing simulation software) and the monitoring system
Education
Bachelor of Engineering - Mechanical Engineering
Universidade Federal De Santa Catarina
Florianópolis, Santa Catarina, Brazil03.2020
Skills
- Large Language Models (LLM)
- Computer Vision
- PyTorch
- Python (Programming Language)
- Deep Learning
- Amazon Web Services (AWS)
- PostgreSQL
- Statistics
- Flutter
- Software Architecture Design
- Mobile Application Development
- Project Planning
Languages
English
Bilingual or Proficient (C2)
German
Advanced (C1)
Portuguese
Timeline
Lead Software Engineer
Inlevel GmbH
11.2022 - Current
Software Engineer
Inlevel GmbH
07.2020 - 11.2022
Research Assistant
Werkzeugmaschinenlabor, WZL der RWTH Aachen
10.2019 - 07.2020
Data Scientist
RUDOLPH
08.2018 - 02.2019
Project Manager
PET - MA
08.2017 - 12.2017
Research Assistant
Werkzeugmaschinenlabor, WZL der RWTH Aachen
08.2016 - 09.2017
Bachelor of Engineering - Mechanical Engineering
Universidade Federal De Santa Catarina