PATTY TECHNOLOGY INDEX · 2026.09

We publish how AI systems are actually engineered.

From model training through evaluation, inference infrastructure, and applied systems, 21 technical monographs document the decisions, methods, validation, and failure boundaries Patty uses in real engagements.

Notes
21
Domains
04
Languages
KR / EN
FIGURE 01Four technical layers connected as one operating system

A capable model is not yet an operable AI system. Training lineage must reach evaluation; evaluated models must run on predictable infrastructure; and inference must enter work through explicit authority and evidence.

  1. 01Design models and data
  2. 02Validate against real work
  3. 03Serve under operating constraints
  4. 04Connect controlled work
A

Technical note

Models & Training

  1. A.01

    Pretraining & CPT

    A CPT discipline spanning corpus diagnosis, mixing, tokenization, distributed runs, checkpoints, and serving acceptance.

    Read note
  2. A.02

    SFT Data Pipelines

    Design source ingestion, schemas, deduplication, contamination control, human audit, calibration, and immutable exports.

    Read note
  3. A.03

    RL Alignment: GRPO & RLHF

    Design task suitability, rollouts, rewards, GRPO/RLHF choice, reward hacking controls, evaluation, and recovery together.

    Read note
  4. A.04

    Training Dataset Generation

    Design source taxonomy, synthetic generation, independent audit, sensitive-data removal, coverage, and human sampling.

    Read note
  5. A.05

    Korean-Specialized Modeling

    Address morphology, endings, spacing, token efficiency, Korean-English mixing, long sessions, and evaluation gaps together.

    Read note
  6. A.06

    Custom Model Development

    Design the build-versus-adapt decision, organization-specific layers, versioning, safety gates, and deployment boundaries.

    Read note
B

Technical note

Evaluation & Benchmarks

  1. B.01

    Benchmark Design & Engineering

    Combine representative workloads, scoring rules, contamination controls, human review, and versioning in one evaluation protocol.

    Read note
  2. B.02

    Model Evaluation & Red Teaming

    Manage capability, safety, security, bias, and tool-use failures through threat models and evidence-based release gates.

    Read note
C

Technical note

Inference & Infrastructure

  1. C.01

    Distributed GPU Serving & Scheduling

    Model request shapes, parallelism, batching, routing, and failure domains together to engineer service levels.

    Read note
  2. C.02

    KV Cache Architecture

    Design block lifecycle, prefix reuse, tiered storage, eviction, and tenant boundaries around the workload.

    Read note
  3. C.03

    MoE Serving (Wide EP)

    Co-design batching, routing, all-to-all communication, and failure scope for wide expert parallel deployments.

    Read note
  4. C.04

    Multimodal Inference

    Engineer resolution, ordering, caching, batching, and output evidence for image, document, and audio inputs.

    Read note
  5. C.05

    RL Rollout Acceleration

    Build a measurable lifecycle across inference workers, sample buffers, reward computation, and trainers.

    Read note
  6. C.06

    GPU Farm Operations

    Unify capacity planning, telemetry, workload isolation, maintenance, and recovery drills in one operating model.

    Read note
D

Technical note

Applied AI

  1. D.01

    Korean-Specialized RAG

    Design the path from Korean document ingestion through structure-preserving chunking, hybrid retrieval, reranking, evidence, and access control.

    Read note
  2. D.02

    Document Intelligence Pipelines

    Preserve layouts, tables, forms, and signatures across HWP, HWPX, PDF, and images while linking extraction, validation, and approval.

    Read note
  3. D.03

    LLM-Native Browser

    Engineer DOM, accessibility, and visual state with session isolation, tool authority, approvals, and reproducible execution records.

    Read note
  4. D.04

    Agent Orchestration

    Design task decomposition, tool contracts, state, approvals, retries, and compensating transactions as an observable graph.

    Read note
  5. D.05

    AI Workforce Transformation

    Connect work decomposition, risk tiers, roles, approvals, quality measurement, and change management in an operating model.

    Read note
  6. D.06

    ViewSpec Rendering

    Place typed component vocabulary, data binding, accessibility rules, policy validation, and visual regression at the rendering boundary.

    Read note
  7. D.07

    On-Premises Deployment

    Design network boundaries, supply chain, secrets, model import, observability, patching, backup, and handover from a threat model.

    Read note

Begin with the bottleneck in your current system.

We map representative workloads, system constraints, and acceptance evidence before proposing an intervention.

Request a technical review