Gasyd GAGNON

- Certified Azure AI 102 - Voice Medical Assistant : Developed a Siri-like assistant, enabling physiotherapists to manage and operate massage machines via voice commands (Langchain, RAG, Speech-to-Text, Text-to-Speech) - Automated Image Generation: Automated SEO-compliant image descriptions using GPT-4 Vision - Legal Bot: Created a RAG-based web app to assist ruminant vets with legal and compliance needs - AI Agent : Developed an AI agent capable of evaluating nutritional content of meals based on images of the meals (Langchain, RAG, Computer Vision) - Deployment and management of AI solutions on Kubernetes using Docker containers, ensuring enhanced scalability and resilience

- Designed and developed interactive dashboards to monitor and communicate key performance indicators (KPIs) to stakeholders and management. - Collaborated with cross-functional teams to identify analytical needs and translate business requirements into data-driven solutions. - Analyzed marketing and business datasets to identify trends, customer behaviors, and optimization opportunities. - Developed and deployed machine learning models for customer segmentation, marketing performance prediction, scoring, and recommendation systems. - Evaluated existing processes and identified opportunities for automation and performance improvements. - Implemented Big Data and streaming data systems to support large-scale data processing and analytics. - Ensured the maintenance, monitoring, and reliability of automated data pipelines and systems. - Managed project deliverables and timelines while maintaining high-quality standards. - Delivered actionable insights and recommendations based on KPI analysis and dashboard results. - Communicated analytical findings and model results through clear visualizations and presentations for business stakeholders.

- Modeling data with DBT or Paradime - Creating and orchestrating data pipelines - Building and training machine learning models - Data visualization using Power BI and Google Data Studio

- Analysis of customer data (usage, top-ups, churn) to identify high-value behaviors and customer segments - Development of churn prediction models (BG/NBD model) to anticipate customer attrition - Advanced customer segmentation using a hybrid approach combining behavioral embeddings and K-means clustering, enabling a more accurate capture of user similarities - Creation of dashboards and visualizations (Power BI) to monitor business KPIs and support decision-making

- Developed efficient NLP and multimodal models by applying model compression, pruning, and knowledge distillation to reduce latency and memory footprint while maintaining performance. - Built explainable AI approaches for language models to analyze attention patterns and interpret model decisions. - Designed multimodal systems combining text and vision for tasks such as semantic search, captioning, and question answering. - Optimized deep learning architectures under compute and deployment constraints for scalable real-world AI systems.

- Built deep question answering systems and machine reading models on multi-source corpora (news, encyclopedias) using encoder–decoder and seq2seq architectures, optimized on SQuAD-style benchmarks. - Developed conversational AI systems with RNN/LSTM/seq2seq, including intent detection, slot filling, and multi-turn context management. - Designed NLU pipelines for intent extraction, named entity recognition (NER), and relation extraction at scale from unstructured text. - Implemented seq2seq models with attention mechanisms to improve contextual memory and natural language generation. - Developed machine reading and knowledge extraction systems for large document collections, enabling multi-document reasoning and information retrieval. - Trained multi-task NLP models (QA, classification, generation) with shared representations and domain-specific fine-tuning.