Building intelligent systems at the intersection of machine learning & real-world impact — from Paris.
About
I'm Ayzel, a Machine Learning Engineer based in Paris, with expertise spanning the full ML lifecycle — from exploratory data analysis and model training to production deployment and LLM-powered applications.
I help companies and startups transform their data assets into tangible AI products. Whether you need a fine-tuned language model, an intelligent search pipeline, or a fully autonomous agent, I design pragmatic solutions that actually ship.
Services
End-to-end ML expertise — from proof-of-concept to scalable production systems.
Design and deploy custom large language model pipelines using OpenAI, Claude, Mistral, and open-source models. Prompt engineering, structured outputs, function calling, and multi-turn dialogue systems.
LLM / NLPRetrieval-Augmented Generation systems that ground your LLM on private data. Full stack: ingestion, embedding, vector search (Qdrant, Pinecone, PGVector), and retrieval-aware generation.
Semantic SearchBuild autonomous agents using LangChain, LlamaIndex, or custom orchestrators. Tool use, memory systems, and multi-agent workflows that automate complex business processes.
Agentic AIFine-tune open-source LLMs (Llama 3, Mistral, Falcon) on your proprietary data with LoRA / QLoRA techniques. Domain adaptation, instruction-following, and RLHF-aligned training.
Fine-TuningDesign robust ML architectures: feature pipelines, model registries, A/B testing frameworks, and monitoring dashboards. Production-ready, observable, and maintainable.
MLOpsCustom deep learning models for object detection, classification, NER, text classification, and multimodal tasks. From data labelling strategy to model deployment.
Deep LearningExploratory analysis, statistical modelling, forecasting, and dashboards. Turn unstructured or siloed data into business insights your team can act on.
AnalyticsStrategic AI roadmap for your organisation. Identify high-ROI automation opportunities, audit existing models, reduce hallucination risk, and define responsible AI policies.
ConsultingFastAPI / Flask backends that serve ML models at scale. Clean REST or streaming APIs, authentication, rate limiting, and cloud deployment on GCP, AWS, or Azure.
Backend / APIProcess
A clear, collaborative process from first call to deployment.
We define the problem, scope, and success metrics. No tech jargon — just honest alignment on what you need and what's feasible.
I build a rapid proof-of-concept, validate assumptions against real data, and surface the risk areas early before committing to full development.
Full development with weekly demos. Clean, documented code, tested pipelines, and continuous feedback loops so nothing ships as a surprise.
Production deployment, monitoring setup, and full knowledge transfer. I don't disappear — post-launch support and maintenance are always available.
Contact
Open for freelance projects, long-term collaborations, and consulting engagements.
Send a message