🏗️ Architecture & Performance

Data Designer is an orchestration framework that coordinates synthetic data generation workflows. It is a client of LLM inference servers—it does not host models itself.

This guide explains the architecture, execution model, and how to tune performance for your specific use case.

Separation of Concerns

┌─────────────────────────────────────┐          ┌─────────────────────────────────────┐
│         Data Designer               │          │       Inference Server(s)           │
│         (Orchestration)             │  HTTP    │       (LLM Hosting)                 │
│                                     │  ─────►  │                                     │
│  • Dataset workflow management      │          │  • Model weights and execution      │
│  • Column dependency resolution     │          │  • GPU allocation and scheduling    │
│  • Batching and parallelism         │          │  • Request queuing                  │
│  • Retry and error handling         │          │  • Token generation                 │
│  • Adaptive concurrency (AIMD)      │          │  • Rate limiting (optional)         │
│  • Data validation and quality      │          │                                     │
└─────────────────────────────────────┘          └─────────────────────────────────────┘
              ▲                                                    ▲
              │                                                    │
        Your workflow                                    Your infrastructure
         configuration                                    (or cloud API)

What Data Designer Does

• Orchestrates the generation workflow across multiple columns
• Resolves dependencies between columns (DAG-based execution)
• Batches work into manageable chunks (buffer_size)
• Parallelizes LLM calls within batches (max_parallel_requests)
• Adapts to rate limits automatically via AIMD concurrency control
• Handles errors with retries and early shutdown logic
• Validates generated data against schemas and constraints

What Data Designer Does NOT Do

• Host models: You must provide LLM endpoints
• Manage GPUs: Your inference server handles GPU allocation
• Scale inference: You must provision sufficient capacity
• Impose rate limits: Your server or API gateway sets rate limits (Data Designer reacts to them automatically)

Execution Model

The async engine processes datasets in row groups, with parallel operations across ready cells and independent columns.

How It Works

Step 1: Split into row groups

Your dataset is divided into row groups of buffer_size records. Row groups checkpoint independently so interrupted runs can resume from durable parquet files.

Step 2: Schedule ready work

Data Designer builds an execution graph from column dependencies. Cells and full-column tasks whose dependencies are satisfied can run concurrently, while order-dependent columns are protected by scheduler locks.

Example workflow:

Row group 1 (100 records)
│
├─► Column 1: category (Sampler)      ──── All 100 values generated
├─► Column 2: prompt (LLM Text)       ──── All 100 values generated
├─► Column 3: response (LLM Text)     ──── All 100 values generated
├─► Column 4: score (Expression)      ──── All 100 values computed
│
└─► Checkpoint row group to disk
    │
    ▼
Row group 2 (100 records)
    ...repeat...

Step 3: Generate cells in parallel

Within each column, cells are processed in parallel up to the configured limit:

Key Concepts

Concurrency Formula

For a given model, async request concurrency is bounded by:

1
concurrent_requests = min(
2
    active_ready_model_cells,  # Ready cells across active row groups
3
    current_admission_limit,   # AIMD-managed limit (≤ max_parallel_requests)
4
    max_in_flight_tasks,       # Global scheduler task-lease ceiling
5
)

active_ready_model_cells depends on the column DAG and the active row-group horizon. Each active row group contributes up to buffer_size rows per ready model column. Other work may consume part of the global max_in_flight_tasks budget. max_parallel_requests sets the per-model ceiling. The actual limit (current_admission_limit) is managed at runtime by an AIMD (Additive Increase / Multiplicative Decrease) request-admission controller that reacts to rate-limit signals from the inference server:

• During optional startup ramp: when startup_ramp_seconds is greater than 0, a new request domain starts at one concurrent request and increases linearly toward max_parallel_requests over that duration.
• On the first 429 in a burst: the limit is reduced by a configurable factor (default: 25% reduction) and a cooldown is applied. Further 429s from already in-flight requests in the same burst do not reduce the limit again — they release their permits and hold the limit steady.
• After consecutive successes: the limit increases by 1 (by default) until it reaches the ceiling or a stabilized rate-limit threshold.

This means Data Designer automatically finds the right concurrency level for your server without manual tuning.

Example: With buffer_size=100 and max_parallel_requests=32, Data Designer can send up to 32 requests in parallel. If startup_ramp_seconds=30, it starts at one request and climbs linearly toward 32 over 30 seconds. If the server returns 429s, startup ramp stops, concurrency drops automatically (e.g., to 24, then 18), and normal AIMD recovery takes over once the server catches up.

Configuration Parameters

Load RunConfig from YAML for one CLI run

data-designer create can load one local .yaml or .yml file with --run-config or -c. The YAML root maps directly to RunConfig fields; do not add a run_config: wrapper.

1
buffer_size: 250
2
max_in_flight_tasks: 128
3
display_tui: false
4
progress_interval: 10.0
5
request_admission:
6
  multiplicative_decrease_factor: 0.75
7
  additive_increase_step: 1
8
  successes_until_increase: 25
9
  cooldown_seconds: 2.0
10
  startup_ramp_seconds: 30.0

$
data-designer create dataset.yaml --run-config run-config.yaml
$
# Equivalent short option; the explicit flag overrides display_tui in the YAML.
$
data-designer create dataset.yaml -c run-config.yaml --no-tui

Partial files are valid. The effective settings use this precedence, from lowest to highest:

1. The active DataDesigner.run_config baseline (the built-in defaults today)
2. Top-level fields explicitly supplied in the YAML
3. Explicit --tui or --no-tui flags

Supplying request_admission replaces that nested baseline object rather than deep-merging it; omitted nested fields use RequestAdmissionTuningConfig defaults. The file is validated through the same RunConfig schema as the Python API, including field constraints, unknown-field rejection, normalization, and deprecated-key compatibility.

buffer_size (RunConfig)

Controls how many records are processed per row group.

1
import data_designer.config as dd
2
from data_designer.interface import DataDesigner
3
4
run_config = dd.RunConfig(buffer_size=2000)
5
6
designer = DataDesigner()
7
designer.set_run_config(run_config)

When to increase: High-capacity inference server, single-model workflows, memory not constrained

When to decrease: Memory-constrained environments, development/debugging, complex multi-model pipelines

max_concurrent_row_groups (RunConfig)

Sets the fixed hard cap on async row groups that may be active at once. The default is 3; each row group contains up to buffer_size records.

1
run_config = dd.RunConfig(
2
    buffer_size=1_000,
3
    max_concurrent_row_groups=8,
4
)
5
designer.set_run_config(run_config)

A wider horizon can expose more ready work to high-capacity endpoints, but it retains more row data in memory and may delay checkpoints. It does not adapt during a run or replace max_parallel_requests, which remains the per-model request ceiling. Treat this as an expert override: benchmark before and after changing it.

Resuming Interrupted Runs

Long generation jobs can be resumed from checkpoints by passing resume to DataDesigner.create() or data-designer create --resume.

1
from data_designer.interface import DataDesigner, ResumeMode
2
3
designer = DataDesigner()
4
results = designer.create(
5
    config_builder,
6
    num_records=10_000,
7
    dataset_name="training_data",
8
    resume=ResumeMode.IF_POSSIBLE,
9
)

Resume modes:

• ResumeMode.NEVER (default): always start a fresh generation run. If the dataset directory already exists, Data Designer writes to a timestamped directory.
• ResumeMode.ALWAYS: resume the existing dataset directory. Raises if the checkpoint is incompatible or cannot be resumed safely.
• ResumeMode.IF_POSSIBLE: resume when the stored config fingerprint matches the current config; otherwise start a fresh timestamped run.

Data Designer does not store the complete effective RunConfig. metadata.json records selected resume identity values, including buffer_size, preserve_dropped_columns, the target record count, and dataset-config identity fields; builder_config.json stores the dataset builder configuration. Resume scans completed batch_*.parquet row groups and reads parquet metadata for the row count actually persisted. That keeps resume crash-safe even if a run was interrupted between writing a row-group parquet and updating metadata, because the filesystem reflects the durable state even when metadata lags by a step.

Resume has a few important invariants:

• buffer_size must match the original run.
• preserve_dropped_columns must match the original run.
• num_records must be at least the original target; you may extend a run by requesting more records.
• Row counts must stay stable within a run. Put filtering, expansion, aggregation, or deduplication at workflow boundaries.
• Once process_after_generation() has run, the dataset is considered terminal for resume. Re-running with the same target returns the existing dataset; extending requires a fresh run.
• If a run crashed after every row group was written but before process_after_generation() could start, resume runs after-generation on the existing on-disk dataset (the parquet files are still clean) and marks it terminal afterwards. A crash during process_after_generation() still raises — the parquet files may have been partially rewritten and starting fresh is the only safe option.

The DatasetCreationResults returned by a resume invocation reflects the full dataset on disk for anything that reads the artifact directory (load_dataset, count_records, load_analysis, export, push_to_hub). Per-run observability — task_traces, model-usage logs, and telemetry events emitted during the call — is scoped to the resume invocation only; the original run’s in-memory traces are not persisted across process boundaries.

max_parallel_requests (InferenceParams)

Sets the maximum concurrent LLM API calls per model. This is the ceiling that the AIMD request-admission controller can ramp up to — the actual concurrency at runtime may be lower if the server signals rate limits.

1
import data_designer.config as dd
2
3
model = dd.ModelConfig(
4
    alias="my-model",
5
    model="nvidia/nemotron-3-nano-30b-a3b",
6
    provider="nvidia",
7
    inference_parameters=dd.ChatCompletionInferenceParams(
8
        max_parallel_requests=8,
9
    ),
10
)

Default: 4

When to increase: Your inference backend has high throughput capacity, you’re using a cloud API with generous rate limits, or you’re running vLLM/TensorRT-LLM with multiple GPUs. With AIMD, setting an aggressively high value is safer than before — the system will self-correct downward if the server can’t keep up, and salvage rounds can reclaim transiently failed rows.

When to decrease: You want to cap resource usage to a known safe level, or you want more predictable/debuggable execution.

non_inference_max_parallel_workers (RunConfig)

This field is declared for non-LLM worker concurrency, but the production engine does not currently consume it.

1
run_config = dd.RunConfig(non_inference_max_parallel_workers=8)
2
designer.set_run_config(run_config)

Default: 4

Changing or loading this value does not change execution today.

Adaptive Request Admission (RunConfig)

Data Designer uses an AIMD (Additive Increase / Multiplicative Decrease) controller to automatically adjust concurrency per model based on rate-limit feedback from the inference server. The defaults work well for most workloads. Override the recovery behavior through request_admission only when you understand the trade-offs.

1
import data_designer.config as dd
2
from data_designer.interface import DataDesigner
3
4
run_config = dd.RunConfig(
5
    request_admission=dd.RequestAdmissionTuningConfig(
6
        multiplicative_decrease_factor=0.75,  # Multiply the limit on a 429
7
        additive_increase_step=1,             # Add slots after each recovery window
8
        successes_until_increase=25,          # Successful releases per recovery window
9
        cooldown_seconds=2.0,                 # Fallback when Retry-After is absent
10
        startup_ramp_seconds=0.0,              # 0 disables the startup ramp
11
    ),
12
)
13
14
designer = DataDesigner()
15
designer.set_run_config(run_config)

Error Handling (RunConfig)

Control retry behavior and early shutdown for failed generations.

1
run_config = dd.RunConfig(
2
    max_conversation_restarts=5,           # Full conversation restarts (default: 5)
3
    max_conversation_correction_steps=0,   # In-conversation corrections (default: 0)
4
    disable_early_shutdown=False,          # Enable early shutdown (default)
5
    shutdown_error_rate=0.5,               # Shut down if >50% errors
6
    shutdown_error_window=10,              # Min tasks before error monitoring
7
)
8
designer.set_run_config(run_config)

When to adjust:

• Strict schemas: Increase max_conversation_restarts to 7, add max_conversation_correction_steps=2
• Debugging: Set disable_early_shutdown=True to see all errors
• Simple text: Reduce max_conversation_restarts to 3

Async Engine

The async engine is the execution path. It dispatches work at the cell level rather than the column level, so independent columns overlap in time and per-(provider, model) AIMD pools tune themselves independently. See the Async All the Way Down dev note for the full architecture.

Per-model timeouts drive every deadline

The inference_parameters.timeout field on a ModelConfig sets the per-request HTTP timeout. The same value also drives the sync→async bridge that custom columns use when they call model.generate(). There is no separate queue-wait deadline — waits scale with provider speed and AIMD’s adaptive concurrency. Slow self-hosted endpoints (e.g. large models on a single GPU) only need this one knob raised:

1
import data_designer.config as dd
2
3
config_builder.add_model_config(
4
    dd.ModelConfig(
5
        alias="slow-model",
6
        model="my/slow-model",
7
        provider="my-provider",
8
        inference_parameters=dd.ChatCompletionInferenceParams(
9
            timeout=600,
10
        ),
11
    )
12
)

Run outcomes

A run can finish with fewer records than requested when non-retryable errors drop rows. Inspect len(result.load_dataset()) to detect.

If the rate of non-retryable errors crosses RunConfig.shutdown_error_rate, generation stops early and raises DataDesignerEarlyShutdownError (a subclass of DataDesignerGenerationError). Catch it separately when a typed retry path is appropriate:

1
from data_designer.interface.errors import DataDesignerEarlyShutdownError
2
3
try:
4
    result = dd_instance.create(config_builder, num_records=1000)
5
except DataDesignerEarlyShutdownError:
6
    # e.g. retry against a different model alias
7
    ...

Local OpenTelemetry Metrics

DataDesigner.create() starts or reuses a process-wide OpenTelemetry Prometheus endpoint on http://127.0.0.1:9464/metrics by default. Use RunConfig to disable metrics for an invocation or choose another port:

1
import data_designer.config as dd
2
from data_designer.interface import DataDesigner
3
4
designer = DataDesigner()
5
designer.set_run_config(dd.RunConfig(otel_metrics_port=None))
6
7
# Or keep metrics enabled on a different loopback port.
8
designer.set_run_config(dd.RunConfig(otel_metrics_port=9465))

Configure a Prometheus server on the same host to scrape the endpoint:

1
scrape_configs:
2
  - job_name: data-designer
3
    metrics_path: /metrics
4
    static_configs:
5
      - targets: ["127.0.0.1:9464"]

The endpoint exposes seven instruments. Prometheus derives throughput and request-completion rates from their counters and histogram counts.

The listener and cumulative metrics remain available until process shutdown so a scrape can collect results after a job finishes. A later enabled job can rebind an idle listener to a different configured port without resetting those metrics. Concurrent jobs that request different ports share the active listener and emit a warning instead of replacing it.

Generated and dropped record counts advance when a row group is durably checkpointed, so data_designer.dataset.progress moves in buffer_size-sized steps. Use data_designer.model.request.active and the gen_ai.client.operation.duration histogram count for live request activity between checkpoints.

Setting otel_metrics_port=None records nothing for that invocation. If an earlier enabled job already started the process listener, disabling a later job does not stop the listener or remove the earlier cumulative metrics.

Prometheus pull is metrics-only: data_designer.log.records counts safe log metadata, not raw log bodies. Existing raw logs continue to use the configured stdout and file handlers and should be collected from those outputs. The Python Prometheus exporter does not support multiprocessing, so this endpoint is unsupported for multiprocessing-based collection.

Common Problems

Tuning Workflow

1. Start with defaults for initial development — AIMD handles rate-limit adaptation automatically
2. Profile your workload: How many LLM columns? How many records? What models?
3. Identify bottleneck: Low GPU util → increase max_parallel_requests (AIMD will self-correct if you overshoot). Memory issues → decrease buffer_size. Long tails → tune retry settings.
4. Check request-admission logs: Look for “concurrency reduced” / “concurrency increased” messages to understand whether rate limits are the bottleneck
5. Iterate: Make one change at a time, measure impact before next change

Related Documentation

• Deployment Options: Library vs. Microservice: Choosing between library and microservice
• Model Configuration: Complete model settings reference
• Inference Parameters: Detailed parameter reference