Industrial data depth × LLM product ownership — the rare pair that robot manufacturing and traditional manufacturing AI both keep failing to find.Two proof cases, one transition: data primitives proven on industrial vehicle fleets (as a team member), and an LLM product I plan, build, deploy, and operate alone — both pointed at robot- manufacturing foundation-model data infrastructure and traditional-manufacturing AI workflows.
Models open up. Data pipelines don’t.RT-2, GR00T, Cosmos, π0 — foundation models keep moving. What blocks the path to production is not the model but the data pipeline. And the people who have built data pipelines rarely overlap with the people who have run LLM products.
Three pillars proven on an industrial vehicle fleet. They port directly into robot manufacturing and traditional manufacturing.
Verified environment — industrial vehicle fleet telemetry pipeline (team work).A 4-tier distributed telemetry system delivered through team work. First-person plural; own contribution stated explicitly.
[vehicle terminal] → Webhook → Bridge InfluxDB → V2InfluxConverterProcess (multi-process) → Measurement InfluxDB → Celery batch → Avro/GCS
| Industrial vehicle fleet | Robot / manufacturing |
|---|---|
| 4-pack BMS async signals per vehicle | 30+ joints + F/T + vision per robot · N machines per line |
| CAN ISO-TP multi-frame | ROS2 chunked / OPC-UA chunked |
| Per-model .dbc / Excel DSL | Per-robot URDF / per-PLC vendor protocol |
Verified environment — untamedai.me, plan → build → deploy → operate, as a solo full-stack.untamedai.me — an AI friend that remembers your feelings. §4 (team / industrial data / constrained disclosure) and §5 (solo / LLM product / open) form a deliberate pair — the contrast itself is the message.
What I want to build — robot-manufacturing and existing-manufacturing AI workflows.Current assets are industrial data pipelines and LLM product operations. Robotics, semiconductor, and steel domain depth are separated honestly as post-hire learning areas.
Teleop demos → automatic builds in RLDS / TFDS / Open X-Embodiment formats. Multi-source time alignment (vision · proprio · action · language) → quality filtering → segmentation → augmentation. Data quality at training time is the model’s ceiling; lifting that ceiling is the pipeline’s job. (deps: P1 + P2 + P3)
Reconciling simulator output vs real-robot telemetry on time, units, and distribution. Domain-randomization parameter distributions sourced from measured data automatically. Reality-gap metric dashboards. Sim-to-real failures are almost always alignment failures. (deps: P2)
Vision-Language-Action triplet sync, mining-ratio control across failure / success, automatic long-horizon segmentation. Splitting language instructions into semantic units + the cost / safety / iteration loop of LLM ops are exactly what untamedai.me handles daily. (deps: P2 + P3 + LLM product ops)
Per-station measurements as a robot traverses the line + post-ship field telemetry, joined causally. End-of-line QC → field-failure traceability as one system. (deps: P2 + P3)
A unified telemetry pipeline across multi-vendor PLC + OPC-UA + MTConnect for semiconductor / steel / cell / display lines. Production ops and model-training data on the same substrate. (deps: P1 + P2 + P3)
Machine-telemetry time-series → predicted end-of-line inspection results. Gradient-boosting baseline → Temporal Fusion Transformer / Patch-TST. Cycle-definition alignment in time is harder than the model itself. (deps: P2)
A natural-language interface for operators — “what caused the line-3 alarm at 02:00 last night?” style RAG over machine logs + SOP + history. Having owned this kind of LLM system end-to-end (§5 untamedai.me) ports directly into the line-assistant problem. (deps: P3 + LLM product ops)
Which machine on the line is the source of the defect? SHAP-based contribution decomposition, drift monitoring, training-distribution guards. (deps: P2 + P3)
The two sub-sections look separate but both run on the same three primitives from §3. That is why the same person ports cleanly into either domain.
The same primitives also work for fleet operations. The first priority is §6 (manufacturing + foundation data); these adjacent areas remain ready to deploy: a unified telemetry substrate across mixed fleets (humanoids / AMRs / cobots) · motor & joint predictive maintenance (RUL regression) · in-operation motion-anomaly detection (autoencoder / GMM). Primitive deps: P1 + P2 + P3 (same as §6).
One data substrate, six AI outlets. People who have only handled the model don’t carry it to production. Only people who have handled the data pipeline and run an LLM product carry it all the way.
| AI workload | Primitive deps | LLM-ops leverage |
|---|---|---|
| Imitation-learning data build | P1 + P2 + P3 | — |
| Sim-to-real telemetry alignment | P2 | — |
| VLA triplet curation | P2 + P3 | ⭐ instruction segmentation |
| Cycle-level quality prediction | P2 | — |
| Time-series anomaly detection | P2 | — |
| Operator LLM assistant (RAG over logs / SOP) | P3 + LLM ops | ⭐⭐ direct 1:1 mapping |
How I work — process signal. From running untamedai.me solo and from team work on industrial data systems, I have learned that how you work matters as much as the result. Three working postures.
Stack used on the industrial vehicle fleet, mapped to the equivalents that port into robot manufacturing and traditional manufacturing. Production code stays under NDA — selective OSS extraction is a later question.
| Ingestion / Bus | Industrial Fleet: Django · Flask webhook → Robot · Mfg: ROS2 · DDS · Kafka · OPC-UA · MQTT |
|---|---|
| Time-series store | Industrial Fleet: InfluxDB → Robot · Mfg: TimescaleDB · ClickHouse · MCAP |
| Metadata DB | Industrial Fleet: MySQL → Robot · Mfg: PostgreSQL |
| Distributed task | Industrial Fleet: Celery + django-celery-beat → Robot · Mfg: Celery · Airflow · Dagster · Ray |
| Process pool | Industrial Fleet: multiprocessing → Robot · Mfg: Ray · Dask |
| Replay format | Industrial Fleet: Avro → Robot · Mfg: MCAP · Parquet · RLDS |
| Storage | Industrial Fleet: GCS → Robot · Mfg: S3 · Azure Blob |
| LLM stack (untamedai.me) | Next.js (frontend) · Python FastAPI (backend) · Supabase (DB) · Cloudflare (hosting) · GPT + Claude Opus (LLM) · Polar (payments) → Foundation-model data / VLA / RAG |
Woon · Industrial Real-Time Data + LLM Product Engineer.Industrial vehicle-fleet telemetry pipeline as a team member → an LLM product (untamedai.me) operated solo → next: robot-manufacturing foundation-model data R&D, or traditional-manufacturing AI workflow pipelines.