AI Full-Stack Engineer | Computer Science Graduate specializing in NLP & Computer Vision
Currently building cutting-edge AI solutions at Nymphar.ai
• Expert in Agentic AI, Modular RAG pipelines, and end-to-end AWS / Azure deployments
Computer Science Engineering graduate with AI specialization in Natural Language Processing & Computer Vision. I build production-ready AI systems from scratch, specializing in:
- Agentic AI architectures for complex reasoning and task orchestration
- Modular RAG pipelines with self-correcting mechanisms, evaluation frameworks, intelligent routing, retrieval fusion, and advanced generation
- End-to-end deployment on AWS / Azure infrastructure (Lambda, ECS, S3, Lightsail or Azure Functions, Container Instances, Blob Storage, App Service)
- Full-stack development from research prototypes to production systems
AI Full-Stack Engineer @ Nymphar.ai (Current)
- Designing and implementing advanced AI systems and LLM-powered applications
Recent Client Work:
- 🏐 Mintonetics - Volleyball data analysis platform with ML-driven insights
- 📧 EmailingData - Klaviyo data analysis and optimization platform
- 🤖 Multiple NLP & RAG-focused missions - Building custom retrieval-augmented generation systems for various clients
- Frameworks: PyTorch, DSPy, LangChain
- Specializations: RAG architectures, Agentic AI, NLP, Computer Vision, Diffusion Models
- Tools: Transformers, LSTM/CNN architectures, Federated Learning
- AWS: Lambda, ECS, S3, Lightsail, full end-to-end deployment
- CI/CD: GitLab CI/CD pipelines, automated testing & deployment
- Orchestration: Docker, Docker Swarm, Kamal
- Infrastructure: AWS VPS management, SSL automation
- Languages: Python, Rust
- Data Processing: Apache Spark (distributed graph computing)
- APIs: RESTful services, Discord bots
- Frameworks: SvelteKit, mdsvex
- Custom Modules: File-system based routing, SSR optimization
Pull Request #822 - Contributed during my internship at Trinity College Dublin while developing a production RAG pipeline for TCD's research infrastructure.
Collection of academic papers I've read and implemented. Bridging theory and practice through hands-on implementation of cutting-edge research.
Respecialization of an InstaFlow-like diffusion model on the Russian Emojis dataset. Explored base model limitations and documented findings on proper deep learning library usage. Fine-tuned from XCLiu/InstaFlow base model.
Comprehensive NLP laboratory assignments covering foundational algorithms to advanced neural network architectures. Complete coursework from my NLP specialization.
Multi-part biomedical ML project: developed ECG classification models using CNN and LSTM architectures, implemented molecule representation and generation via PyTorch Geometric, integrated BioPython for genomics, performed TF binding site classification with CNNs, and explored federated learning approaches.
Large-scale graph processing using Apache Spark. Successfully implemented distributed clique calculation across Spark Master and Worker Nodes for very large graphs.
Forked and fixed Bluetooth communication project to support Master's students in Mechanical Engineering at Trinity College Dublin during my internship.
Complete pipeline for translating French audio to English while preserving the original speaker's voice. Multi-stage processing including speech recognition, translation, and voice cloning.
File-system based TTRPG wiki built from scratch with custom SvelteKit modules:
- Custom file-system routing where folder structure = navigation
- mdsvex integration for enhanced markdown
- Co-located assets and dynamic component loading
- FoundryVTT deployment for D&D virtual tabletop sessions
- Full AWS deployment with Kamal orchestration, SSL automation, and GitLab Container Registry
- Infrastructure: VPS management, Docker Swarm, automated CI/CD
Built complete Discord bot infrastructure in both Python and Rust:
- Full CI/CD pipeline using GitLab CI/CD
- Docker Swarm deployment on AWS Lightsail
- Project runner setup for automated testing and deployment
- YADB (Yet Another Discord Bot) architecture
Custom matching algorithm to optimally assign students to project preferences with complex constraints. Built as class representative to solve real organizational challenges in my engineering school.
- Discord: fiwex
- LinkedIn: linkedin.com/in/adrien-giget
- Website: grimoire.fiwex.dev