AetherGate — LLM Stack
Self-hosted AI infrastructure: Open WebUI + LiteLLM gateway + admin console, Authentik SSO, and Postgres. Local & cloud LLMs ready in about 60 seconds with Docker Compose.
Vikou C Nelson — Full-Stack Developer | AI Builder | LLM Engineer. PhD in Statistics & Econometrics.
What I do
From academic research to production applications — a rigorous, results-driven approach.
RAG chatbots, automated customer support, Telegram assistants, React + FastAPI apps, and self-hosted LLM stacks with Docker.
AutoML (TPOT), prediction pipelines, Kaggle notebooks, exploration and predictive modeling.
Tobit models, censored data, mathematical foundations for rigorous machine learning.
Tech stack
Portfolio GitHub
Deployable applications, intelligent agents, and research work — each project solves a real problem.
Self-hosted AI infrastructure: Open WebUI + LiteLLM gateway + admin console, Authentik SSO, and Postgres. Local & cloud LLMs ready in about 60 seconds with Docker Compose.
Automates 70–80% of support tickets: real-time replies, confidence scores, source citations, automatic escalation to a human. Full RAG stack (ChromaDB, LangChain, GPT-4o).
Generates plans, checklists, and ideas in seconds with GPT-4. React UI + FastAPI backend, one-click Docker deployment.
Dr. Émile — conversational assistant with per-user memory, clinical persona, and GPT-4o-mini. Aiogram 3 + Docker.
Econometric model for censored data — MLE, analytical gradient, scikit-learn API, validated vs R censReg.
Automatic regression pipeline search, Python export, DevContainer / Codespaces, reproducible notebooks.
Data science experiments, analyses, and ML prototyping. Complemented by Kaggle competitions.
Structured notebooks: linear algebra, probability, optimization — solid foundations for machine learning.
AI agent, ML pipeline, statistical modeling, or full-stack app — I reply within 48 hours with a concrete first step.
nvikou@gmail.comBackground
I'm Vikou C Nelson, Full-Stack Developer | AI Builder | LLM Engineer with a PhD in Statistics & Econometrics. I work across AI/ML development, GPT engineering, and full-stack delivery — with a dual strength: mathematical rigor as a researcher and shipping applications ready for production.
I've built self-hosted LLM infrastructure (AetherGate), customer support agents (RAG + human escalation), Telegram conversational assistants, NeuroHelper tools, Tobit models validated against R, and reproducible AutoML pipelines with TPOT.
What drives me: turning your data and ideas into tools that save time, cut costs, and improve decisions. Every collaboration starts by understanding your business problem — not stacking buzzwords.
Contact
Freelance work, technical collaboration, data/AI consulting, or a simple chat — I'd be happy to connect.