About this Garden

#about #profile

🏠 Home · 📚 Reading Notes · 📝 Articles

What this is

This garden is my public thinking space. It’s where I work things out in the open, follow questions that pull at me, and shape ideas into something clearer. Notes here change as I do. Some grow, some get replaced, some get refactored into better versions. Curiosity leads; rigor keeps me honest.

Who I am

sethrobins.jpg|220

I’m an autistic systems thinker who learns by taking things apart and seeing how they fit together. I’ve always been a builder; wiring, fixing, coding, printing, testing. Making things helps me understand the world, and naming the parts helps me explain it.
Most of what I publish here begins as raw curiosity and becomes something structured: a method, a definition, a model, or a story.

What I’m exploring

  • Recursive Distinction Dynamics (RDD). A working theory about how reality builds itself from differences. Draw a boundary, let it recurse, and watch time, coherence, meaning, and agency emerge. This is still early work — a mix of philosophy, information theory, and testable structure.
  • PKM as publishing. Turning private notes into public artifacts, complete with provenance and versioning, so thinking isn’t just stored but improved.
  • Agentic tools. Lightweight, auditable workflows that take real work off my plate. Not hype, practical tools that actually reduce effort.
  • Reading & review. Brief, falsifiable summaries that sharpen understanding and feed back into projects, writing, and research.

Fiction in progress

  • Wake (near-future SF). Humanity finds ancient, indifferent star vessels. We live in their wake like dolphins pacing cruise ships—fascination without contact. The story follows what people build in the shadow of indifference.
  • Devil’s Backbone (horror). An information-parasitic entity—Ixchel—propagates like a recursive algorithm through ritual, worship, and modern platforms. It exploits broken error-correction in human groups. Science, cognition, and belief are the terrain.

What I do (now)

I work on the frontier between industrial operations and intelligent systems. My focus is bringing AI, advanced process control, and decision models into real industrial manufacturing: safely, reliably, and with measurable impact.
I enjoy problems that sit at the intersection of physics, human judgment, and computation. When these systems work well, you don’t just improve a process; you make people’s work lives better, calmer, and more predictable.

Impact snapshot

  • Production-grade deployments inside operating units (not pilots)
  • Measured in plant outcomes: ↑ throughput, ↑ stability, ↓ energy intensity
  • Documented multi-million operational improvement within the last year
  • Early-majority adoption across production, maintenance, engineering, commercial

Operating principles

  • Safety and control are non-negotiable. Architect for fail-safe behavior.
  • Machines monitor and execute precision; humans keep judgment.
  • Cost-aware models. Add complexity only when it pays for itself in performance.
  • Tight feedback loops. Build for continuous measurement and iteration.

Core stack & tooling

Industrial/controls. DCS/PLC/SCADA (MPC/APC hooks, interlocks, ISA-88/ISA-18.2), historians & events (PI/AF, Canary, PHD), OPC UA/DA, Modbus/TCP, MQTT Sparkplug B, Kepware/UA gateways, PI Vision / Ignition HMI, alarm hygiene & MOC.

Data & plumbing. Python/Node on WSL2; pandas/NumPy/statsmodels; SQLite, Postgres, DuckDB; Parquet/Arrow; Kafka/Redpanda, MQTT brokers; lightweight ETL; cost/constraint models for ops metrics.

ML & optimization. scikit-learn, XGBoost; feature stores (lite); OR-Tools/PuLP (LP/MIP); time-series forecasting & variance reduction; model monitoring in production contexts.

LLMs & agentic systems. Prompt design & tool-use/function calling, structured outputs (JSON Schema/Pydantic), RAG (FAISS/Chroma/pgvector), LangChain/LlamaIndex when useful, small custom agents (policies, evaluators, planners), evals/guardrails (RAGAS-style checks, deterministic parsers), observability (Langfuse/W&B).

Orchestration & CI. Node-RED, n8n; Airflow/Prefect; GitHub Actions/Jenkins; Task Scheduler/systemd; FastAPI/Flask microservices; Docker/Compose for edge deploys.

Cloud & infra. AWS (S3, Lambda, Step Functions, ECS/Fargate, IoT Core/Greengrass, Timestream); VPC/VPN/PrivateLink; secrets management; Grafana/Prometheus/CloudWatch.

Dev & publishing. VS Code, GitHub, Jupyter, Obsidian → Netlify static publishing; dashboards for control-room consumption.

Principle. Machines handle precision & monitoring; humans keep higher-order judgment—safety and control are non-negotiable.

Selected initiatives (condensed)

  • Intelligent control & optimization inside high-margin chemical production
  • Variance reduction and throughput/energy improvements via APC + ML
  • Mobile, intrinsically safe operations (smart devices, plant Wi-Fi, connected workflows) to prepare for intelligence-driven plants
  • Capability development for control, analyzers, and power—training aligned to live risk controls and standards

Roles (short)

  • AI Adoption Consultant — design + deploy intelligent control/optimization in live production
  • Industrial Mobility & Intelligent Ops Architect — intrinsically safe mobile operations, connected worker apps
  • Process Control Capability Development — site-wide learning programs for control/automation disciplines
  • MES / Data Asset Engineer — historian → asset management, reliability and efficiency projects
  • Earlier: process control engineering, instrumentation, and instruction (hands-on to systems)

Contact & elsewhere

🏠 Home · 📚 Reading Notes · 📝 Articles