10/28/2024 updated


100 % available
ML Engineer - Data Science Web - Developer
Funchal, Portugal
Only remote
Master in Computer ScienceInnovationTraining DataArtificial IntelligenceContent ManagementData ManagementDistillationDevOpsWeb DevelopmentDjangoPerformance ManagementPython (Programming Language)Machine LearningNatural Language ProcessingPredictive AnalyticsRecommender Systems
Cognitiva is a dynamic company focused on developing innovative AI solutions. In my role, I engaged with a variety of tasks that positioned me at the forefront of machine learning and web development.
- Developed and optimized machine learning models for the minuva.com platform, with a focus on efficiency for low-resource hardware using ONNX and model distillation.
- Led the deployment and DevOps operations on Google Cloud, ensuring seamless integration and functionality of applications.
- Created an analytics platform for visualizing large language model (LLM) conversations, enhancing user insights.
- Designed and implemented a predictive market platform using Django and Htmx for effective data management and user interaction.
- Developed natural language processing models for conversation taxonomy discovery and implemented innovative methods for streaming data ingestion.
- Improved existing recommendation systems by addressing bias in training data, achieving a notable 28% improvement in performance metrics.
- Enhancing a Content Management System (CMS) for the PortalBitcoin project using Python, Django, and Wagtail. Successfully deploying the website on Google Cloud platform, ensuring a reliable and scalable web presence.
- Designed and developed a predictive market platform that enhanced client decision-making processes.
- Developed and evaluated methods for extracting customer intents from user dialogues utilizing Transformer NLP models and rule-based techniques.
- Implemented and refined a novel dialog segmentation method for customer/agent interactions, which facilitated the creation of a dataset for training an NLP model that effectively segments dialogues.
- Implemented and refined a novel dialog segmentation method for customer/agent interactions, which facilitated the creation of a dataset for training an NLP model that effectively segments dialogues.
- Trained a targeted NLP model to identify user goals from customer/agent dialogues, achieving high performance even with a limited dataset.
- Designed and executed a method to extract salient spans from dialogues, enabling the creation of a dataset aimed at training an NLP model to identify relevant dialogue segments.
Experience
Professional experience0 - 5 years
Freelancer since2020
Languages
EnglishFluentPortugueseNative speaker
Project history
Developed machine learning models for minuva.com platform, optimized models for low-resource hardware, responsible for deployment and DevOps on Google Cloud, developed analytics platform for LLM conversations.
Designed and developed a predictive market platform, implemented backend in Django and frontend in Htmx, responsible for deployment and DevOps on Google Cloud.
Developed a platform for conversation data visualization based on posthog open source project, developed NLP models for conversation taxonomy discovery, developed pipelines for streaming data ingestion in posthog.