I build reliable AI solutions with RAG and LLMOps practices: measurable quality (RAGAS), observability (Langfuse/Helicone) and cost/latency control (serving & caching). Remote-first, async-friendly.
• LLM Integration — Hands-on experience with prompt design, context strategies, basic RAG implementation
• Development Tooling — Docker environments, MCP server configuration, CLI development (NexoCLI)
• Business-Technical Bridge — Translating stakeholder requirements into technical architectures and delivery plans
AI/ML (Hands-on): LLMs, prompt design, context strategies, embeddings, MCP server configuration
Development (Operational): Docker, Python (automation/POC), Git/GitHub, CLI development
LLMOps (Learning): RAGAS evaluations, Langfuse/Helicone traces, performance optimization concepts
Data & Infrastructure: SQL/NoSQL basics, vector databases (foundational), hybrid retrieval concepts
Cloud Platforms: Azure (certified), AWS/GCP foundations
Enterprise Bridge: Power Platform, SAP Build Apps, solution architecture, stakeholder communication
🚀 Featured — NexoCLI
A lightweight CLI to automate developer workflows with pragmatic AI. Designed for reproducibility, short feedback loops, and minimal friction.
- Why it matters: faster scaffolding, consistent outputs, and traceable runs
- What it shows: engineering for reality — metrics, logs, and limits
- Links:
- Repo · NexoCLI_BaseGemini → https://github.com/nsalvacao/NexoCLI_BaseGemini
- Docs/Demo page → https://nsalvacao.github.io/NexoCLI_BaseGemini/
🤖 Featured — Nexo-Agents
A comprehensive collection of 44 production-ready AI agent commands for Google AI CLI integration. Specialized agents for engineering, marketing, design, and operations workflows with automatic context discovery.
- Why it matters: role-specific expertise, structured automation, and native CLI integration
- What it shows: production-grade command design — dynamic parsing, shell integration, and composable workflows
- Links:
- Repo · Nexo-Agents → https://github.com/nsalvacao/Nexo-Agents
- Docs/Demo page → https://nsalvacao.github.io/Nexo-Agents/
- LLMOps & Evals: prompt regression & RAGAS harnesses, Langfuse traces
- RAG & Search: hybrid retrieval (BM25 + dense + rerankers), pgvector/Qdrant
- Serving & Cost: vLLM/TGI comparisons, INT4 vs FP16, batch/speculation
- MCP & Agents: safe read-only tools for triage, observability and CI
| Category | Technologies | Level |
|---|---|---|
| AI/ML | LLMs, Prompt Design, MCP | Hands-on |
| Development | Python, Docker, CLI Tools | Operational |
| LLMOps | RAGAS, Langfuse/Helicone | Learning |
| Data | SQL/NoSQL, Vector DBs | Foundational |
| Cloud | Azure (certified), AWS, GCP | Foundations |
| Enterprise | Power Platform, Solution Architecture | Experienced |
I run compact workshops for teams (3–4h):
- “Reliable RAG in production” — evals, observability, and cost control
- “LLMOps quickstart” — traces, dashboards, and prompt regression basics Open to corporate training & public cohorts.
- Website/Portfolio → https://en.nunosalvacao.pro/
- LinkedIn → https://www.linkedin.com/in/nsalvacao/
- Email → nuno.salvacao@gmail.com
CET (EQF Level 5, PT). Unavailable during internship Sep–Dec 2025 (400h, funded; contract-free); open to opportunities thereafter.