Rafael COELHO

/ Work

Selected projects

Case studies of production ML systems I've designed, shipped, and operated. Architecture diagrams, tradeoff notes, and measurable outcomes.

COELHO Nexus logo

COELHO Nexus

Flagship
JUL 2026

Kubernetes-native multi-agent AI platform hosting three production features — documentation distillation, YouTube GraphRAG search, and autonomous research digests — on one shared LLM rotator, six data stores, and full GitOps observability, architected and operated solo.

Three AI features live on one platform · 124 REST endpoints across four domains · 18 Terraform modules across a 6-layer dependency graph · 20+ free-tier LLMs routed by a variance-aware Thompson-sampling bandit · dual observability (Grafana LGTM + LangFuse) on every call.
Kubernetes Terraform ArgoCD FastAPI FastHTML FastMCP +12 more
COELHO RealTime logo

COELHO RealTime

Flagship
JAN 2026

Production-grade Real-Time MLOps platform on Kubernetes combining incremental ML and batch ML across fraud detection, ETA prediction, and customer segmentation.

39 FastAPI endpoints across three versioned routers · sub-millisecond Redis-cached inference · end-to-end GitOps (Terraform → Helm → ArgoCD → GitLab CI) · full observability stack (Prometheus + Grafana + Alertmanager + Karma).
Kubernetes k3d Kafka Spark Delta Lake MinIO +16 more
YouTube Content Search logo

YouTube Content Search

MAR 2025

Multi-mode GraphRAG platform that extracts entities and relationships from YouTube transcripts into a Neo4j knowledge graph for multi-hop reasoning — the research that became COELHO Nexus.

Four search modes (Search · Video · Channel · Playlist) · multi-hop reasoning via Neo4j knowledge graph · multi-provider LLM routing across Groq, OpenAI, SambaNova, Scaleway.
LangChain LangGraph Neo4j GraphRAG Llama 3.1 Gemma 2 +6 more
COELHO Agents logo

COELHO Agents

FEB 2025

Multi-agent platform combining a memory-aware assistant, an autonomous Software Developer agent with self-fixing code loops, and a GraphRAG-based YouTube Content Search — all orchestrated with LangChain + LangGraph.

Three production agent tools shipped end-to-end · autonomous code-fix retry loops in the Software Developer · GraphRAG (Neo4j) backing multi-hop video query reasoning.
LangChain LangGraph Neo4j GraphRAG Streamlit Python +2 more
COELHO GenAI logo

COELHO GenAI

SEP 2024

Privacy-first Generative AI platform connecting users to open-source LLMs (Llama 3.1, Gemma 2) via Ollama or Groq — with five purpose-built tools spanning chat, retrieval, autonomous data science, document analysis, and agentic plan-and-solve.

Five distinct GenAI tools shipped end-to-end · runs fully local on Ollama or low-latency via Groq · autonomous data-science agents and document analysis via Docling.
LangChain Ollama Groq Llama 3.1 Gemma 2 Docling +5 more
Pentesting Automation logo

Pentesting Automation

JUL 2024

Web vulnerability scanner pipeline covering 23 classes of vulnerability — orchestrates 39 open-source bug-bounty tools (Nmap, Nikto, Nuclei, the ProjectDiscovery suite, and more) through a hand-built dependency graph on Kali Linux to compress hours of manual setup into a single automated pass.

23 vulnerability classes scanned automatically · Go + Bash + Python orchestration on Kali Linux · normalized findings surface highest-signal issues first, freeing human attention for cases that need it.
Go Bash Python Kali Linux NMAP Nikto +4 more
COELHO VISION logo

COELHO VISION

APR 2024

Computer Vision platform combining OpenCV, Ultralytics, and MediaPipe behind four production pipelines — object detection, image segmentation, pose estimation, and real-time camera integration — exposed via a Streamlit interface.

Four production CV pipelines shipped end-to-end · live Streamlit demo · backbone choices (OpenCV / Ultralytics / MediaPipe) picked per pipeline by accuracy-vs-latency tradeoff.
OpenCV Ultralytics YOLO MediaPipe RoboFlow OpenVINO Python +1 more
Formula 1 Analytics logo

Formula 1 Analytics

NOV 2023

F1 analytics platform combining historical narrative, statistical insights, full season reviews, and an AI Space applying ML to race data — running live on Streamlit with end-to-end data pipelines.

Four-section live platform · public Streamlit app · ML on race outcomes + pit strategy modeling · sandbox that doubled as my testbed for sports-analytics tooling.
Python Streamlit pandas NumPy scikit-learn Plotly +2 more
COELHO Finance logo

COELHO Finance

SEP 2023

Financial market analytics platform combining Data Science, ML, and AI applied to investment analysis, portfolio tracking, and opportunity identification — running live on Streamlit with free public access.

Live Streamlit app with end-to-end financial data pipelines · ML models for investment decisioning · the deliberate sandbox where I prototyped frameworks before bringing them into production work.
Python Streamlit pandas NumPy scikit-learn Plotly +2 more