> whoami
Systems-focused engineer who lives in the terminal. I build backend services in Go, provision infrastructure with Terraform, and monitor everything with Datadog. My daily driver is nvim on Arch (btw), window-managed by dwm. I also build agentic AI systems — multi-agent orchestration, autonomous workflows, and LLM tooling that actually does things.
Passionate about observability, monitoring, and making complex distributed systems understandable. Every tool I build follows one rule: do one thing well.
Suckless • Unix • KISS • Agentic
> ls projects/
filePick
TUI file picker with fuzzy search, preview pane, and vim-like keybindings.
github.com/Nokodoko/filePickcomputeCommander
CLI compute orchestration tool for managing cloud instances and local VMs.
github.com/Nokodoko/computeCommanderRayne
Go-based REST API wrapping Datadog for monitoring demos and webhook automation.
github.com/Nokodoko/rayneMessages TUI
Terminal UI for Google Messages with vim-like navigation, QR pairing, and external editor support.
github.com/Nokodoko/messages_tuiopenbrain
AI-powered second brain with LLM classification, pgvector semantic search, Redis caching, and n8n workflow automation.
github.com/Nokodoko/openbrainicarus
Agentic harness for driving fleets of coding agents — multiplexed terminal panes, resident local models, browser automation, and a context layer that persists across sessions.
Private repo — open-sourcing soonMonty
Local AI agent powered by Ollama with persistent memory, tool-calling, and multi-model support.
Coming Soon - Private Repo> cat stack.yml
> cat proof_of_work.md
ECCO Select Helps Drive 60% MTTR Reduction for USDA Forest Service
> Role: Site Reliability Engineer & Observability SME at ECCO Select, embedded with USDA Forest Service
Led the rollout of a full-stack Datadog observability platform across 1,000+ VMs and 150+ mission-critical applications on AWS (EC2, ECS, Fargate). Standardized monitoring deployment with Terraform, integrated Datadog with ITSM for automated ticketing, and built dashboards serving teams at different stages of cloud adoption.
“We used to spend hours figuring out where to route an alert. Now Datadog talks to our ITSM, and the ticket goes straight to the right team.”
> less resume.pdf
Systems Hacker and Agentic AI Builder. Deep expertise in DevOps, SRE, and observability (Datadog), with production Kubernetes and Terraform at enterprise scale. I write backend services and CLI tools in Go, build multi-agent AI systems, and follow the suckless/Unix philosophy. Proven track record delivering monitoring solutions for federal agencies (USDA Forest Service) and private sector. Currently building autonomous agent frameworks, LLM tooling, and observable infrastructure.
> gh contribution-graph
> ls writing/
We Built an AI That Learns From Every Incident
Rayne — a Go API gateway that turns Datadog webhooks into AI-driven RCAs with vector-stored institutional memory and auto-generated investigation notebooks.
Read on LinkedIn →Teaching Your Agent to Know What It Doesn't Know: Building a Live Metacognition Loop
An auto-didact rubric system — every session emits training signals, an RL loop proposes weight deltas gated by a frozen anti-regression corpus, and the updated rubric injects back into the next prompt.
Read on LinkedIn →One-Shot Context: Replacing Keyword Routing with a Six-Signal Pipeline
Precision-at-k=1 context routing: semantic similarity, TrustGraph entity boost, keyword fallback, dual-floor admission gate, and graph-level veto — under a 250ms budget.
Read on LinkedIn →Stop Sending Every Token to Your Most Expensive Model
MoE-inspired routing architecture — local agents handle 80% of tokens while frontier models focus on what matters, with caching layers that cut redundant API calls.
Read on LinkedIn →The Invisible Spend Problem
Tracking costs in multi-agent AI workflows where spend is distributed across parallel processes with no single point of visibility — a four-layer instrumentation system for per-agent cost attribution.
Read on LinkedIn →How I Made AI Write Its Own Requirements — And Then Argue With Itself Until They're Right
A spec-review loop where AI agents draft requirements, then argue with each other until the spec is airtight — before a single line of code gets written.
Read on LinkedIn →The Missing Layer Between You and Your AI Agents
Intent engineering as the control plane between human prompts and autonomous AI agent execution — bridging the gap where instructions get lost in translation.
Read on LinkedIn →Your AI Agents Forget Everything Between Sessions — Mine Don't
OpenBrain Smart Router — dual-hook session lifecycle integration with a 5-stage Sorter/Bouncer/Router/Linker/Compactor pipeline that injects cross-agent memory under a 500ms budget.
Read on LinkedIn →My AI Agents Have a Mail System
Inter-agent communication with SQLite mailboxes — 11 message types, 4 priority levels, broadcast groups, and near-instant delivery via fsnotify
Read on LinkedIn →I Built 42 Hooks That Think Before My AI Acts
Intent engineering pipeline with 42 hooks that verify AI agent actions before execution — a safety layer built on top of Claude Code to prevent agentic mistakes at scale.
Read on LinkedIn →