Drive Faster Analytics with AI Code Generation

A business user wants to see the top-10 sales by category for the last week. The traditional path requires a request to an analyst, refinement, and waiting in queue. The answer comes back a couple of days later. Text-to-Code does it differently. An LLM turns the question into Python or SQL code. Then it executes the code and returns the result with visualization. All this takes seconds. According to Gartner, the share of NLQ queries will grow to 60% in the coming years. Our AI data analysis solution also provides LLM dashboards and NLP data queries for seamless interaction.

Recently, we deployed Text-to-Code for a retail chain with 500 stores. Previously, a request like 'show the average check by region for the past month' took 2 days. Now the answer comes in 15 seconds. Analysts shifted to complex tasks, and business users got self-service. With over 5 years of experience and 50+ AI analytics projects delivered, we ensure reliable implementation. Our solution handles up to 1000 queries per day with average latency under 2 seconds. Implementation costs start at $15,000 and can save up to $100,000 annually in analyst time. Our free audit is valued at $2,000.

How Text-to-Code Accelerates Data Work

Classical BI requires pre-designed dashboards. Each new question takes 1–3 days for approval and development. Text-to-Code cuts this to 10–30 seconds. Analysts spend 70% less time on routine tasks. Business users get self-service for 80% of standard queries. Text-to-Code is 10–50 times faster than traditional BI queries. Typical daily query volume is 500-1000, with peak loads up to 2000.

Why Code Security Is the Main Risk?

The main risk of Text-to-Code is malicious or incorrect code. Our isolation builds on three levels:

• Sandbox container: execution in an isolated environment with restricted file system and network access. We use Docker or gVisor.
• Module whitelist: only pandas, numpy, plotly, and built-in Python functions are allowed. Import requests for third-party libraries are blocked.
• Result validation: output data types are checked, code is logged for audit.

Example Docker container configuration for code isolation:

version: '3.8'
services:
  sandbox:
    image: python:3.11-slim
    command: tail -f /dev/null
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    volumes:
      - ./data:/data:ro
    environment:
      - PYTHONDONTWRITEBYTECODE=1
    deploy:
      resources:
        limits:
          cpus: '1'
          memory: 2G

Implementation Using AIDataAnalyst

from anthropic import Anthropic
import pandas as pd
import io

class AIDataAnalyst:
    def __init__(self, dataframes: dict[str, pd.DataFrame]):
        self.dfs = dataframes
        self.llm = Anthropic()
        self.schema = self._build_schema()

    def _build_schema(self) -> str:
        schema_parts = []
        for name, df in self.dfs.items():
            schema_parts.append(f"Table: {name}")
            schema_parts.append(f"Shape: {df.shape[0]} rows x {df.shape[1]} columns")
            schema_parts.append("Columns:")
            for col in df.columns:
                dtype = str(df[col].dtype)
                n_unique = df[col].nunique()
                sample = str(df[col].dropna().head(3).tolist())
                schema_parts.append(f"  - {col} ({dtype}, {n_unique} unique): {sample}")
            schema_parts.append("")
        return '\n'.join(schema_parts)

    def analyze(self, question: str) -> dict:
        """Analyze data based on a natural language question"""
        system_prompt = f"""You are a data analyst. You have access to these dataframes:
{self.schema}

Write Python code using pandas to answer the user's question.
The dataframes are available as: {list(self.dfs.keys())}
Return ONLY the Python code, no explanations. Use variable 'result' for the final result."""

        response = self.llm.messages.create(
            model="claude-3-5-sonnet-20241022",
            max_tokens=1000,
            system=system_prompt,
            messages=[{"role": "user", "content": question}]
        )

        code = response.content[0].text.strip()
        if code.startswith("```python"):
            code = code[9:-3].strip()

        result = self._execute_safely(code)

        # Generate explanation
        explanation = self._generate_explanation(question, result, code)

        return {
            'question': question,
            'code': code,
            'result': result,
            'explanation': explanation
        }

    def _execute_safely(self, code: str) -> any:
        """Safe execution of generated code"""
        import builtins

        # Allowed functions
        safe_globals = {
            '__builtins__': {
                'len': builtins.len, 'range': builtins.range,
                'list': builtins.list, 'dict': builtins.dict,
                'str': builtins.str, 'int': builtins.int,
                'float': builtins.float, 'print': builtins.print,
                'sorted': builtins.sorted, 'sum': builtins.sum,
                'min': builtins.min, 'max': builtins.max,
                'round': builtins.round, 'abs': builtins.abs,
            },
            'pd': pd,
            'np': __import__('numpy'),
        }

        # Add dataframes
        safe_globals.update(self.dfs)

        local_vars = {}
        exec(code, safe_globals, local_vars)

        return local_vars.get('result')

    def _generate_explanation(self, question: str, result, code: str) -> str:
        result_str = str(result)[:2000] if result is not None else "No result"

        response = self.llm.messages.create(
            model="claude-3-5-sonnet-20241022",
            max_tokens=300,
            messages=[{
                "role": "user",
                "content": f"""Question: {question}
Analysis result: {result_str}

Provide a clear 2-3 sentence business explanation of this result."""
            }]
        )
        return response.content[0].text

How Automatic Visualization Chooses a Chart?

class AutoVisualizer:
    def create_chart(self, data, question: str) -> str:
        """Automatic chart selection and creation"""
        chart_type = self._suggest_chart_type(data, question)

        import plotly.express as px

        if isinstance(data, pd.DataFrame):
            if chart_type == 'bar':
                fig = px.bar(data, x=data.columns[0], y=data.columns[1],
                             title=question[:80])
            elif chart_type == 'line':
                fig = px.line(data, x=data.columns[0], y=data.columns[1:],
                              title=question[:80])
            elif chart_type == 'scatter':
                fig = px.scatter(data, x=data.columns[0], y=data.columns[1],
                                 title=question[:80])
            elif chart_type == 'pie':
                fig = px.pie(data, names=data.columns[0], values=data.columns[1],
                             title=question[:80])

            return fig.to_html(include_plotlyjs='cdn', full_html=False)

        return None

What's Included in Developing an AI System?

We provide a complete set of documentation and artifacts:

• Model card: specification of the selected LLM, inference parameters, library versions;
• Configuration files: Docker Compose, environment variables, deployment scripts;
• Interactive Playbook: description of all components and setup instructions;
• Load testing: report with p50/p99 latency, FLOPS, GPU utilization under peak loads;
• Team training: 2–3 sessions on operation and model fine-tuning;
• Post-release support: 30 days of incident management and improvements.

Model comparison for Text-to-Code:

Work Stages and Estimated Timing

1. Analytics (2–3 days): we examine your data, identify typical queries, select LLM and architecture.
2. Design (3–5 days): RAG schema, sandbox, code-generation pipeline.
3. Implementation (7–10 days): integrate LLM, write components, visualizations.
4. Testing (3–5 days): unit tests, load testing, security checks.
5. Deployment (2–3 days): deploy on your infrastructure or cloud, handover documentation.

Estimated timeline: 17 to 26 days. Cost is calculated individually after auditing your data and requirements. We guarantee code execution safety and data confidentiality. Our LLM analytics platform enables natural language queries. It turns NLP data queries into executable code. The AI BI system integrates seamlessly with existing databases. For MLOps deployment, we use Docker and Kubernetes. We ensure secure code generation through sandbox isolation. This AI BI system supports MLOps deployment. Get a consultation for a project evaluation. Order a free audit of your data.

Data Engineering for ML: Pipelines, Labeling, and Data Quality

“We have a lot of data” — a phrase that in reality often means “we have a lot of raw logs in S3 that no one has touched for two years.” Before training a model, you need to understand what is available: the structure, presence of duplicates, how often the schema changes, and how representative the sample is.

Data Engineering for ML is not just ETL. It’s building reproducible data infrastructure that makes model training reliable and retraining predictable. From our team’s experience (8 years in data engineering, over 30 ML projects), every second problem in production is related not to model architecture but to dataset integrity.

How Are ETL Pipelines for ML Different from BI?

ETL for analytics and ETL for ML are different tasks. Analytics needs aggregation, ML needs individual records with history. Analytics doesn’t require train/val/test split, ML does. Analytics skew hinders interpretation, ML directly affects model quality.

Tools. Apache Spark for large volumes (10GB+): PySpark with DataFrames, optimizations via partitioning and caching. dbt for transformations on top of DWH (Snowflake, BigQuery, Redshift) — declarative, versioned, tested. Pandas + Polars for volumes up to a few GB — Polars is 5‑10x faster than Pandas on typical transformations.

Temporal splits. For ML it’s important that the split is by time, not random. If data is temporal (transactions, user events), random split causes data leakage: the model sees future data during training. Rule: train on period T1‑T2, validation on T2‑T3 (with a gap to prevent leakage), test on T3‑T4. An incorrect split can cost 10–15% of model quality on validation.

Incremental pipelines. The model is retrained weekly on new data. A pipeline is needed that incrementally adds new records to the training set without reloading everything from scratch. Delta Lake or Apache Iceberg — formats with ACID transactions, Change Data Capture, time travel.

What Causes Training‑Serving Skew and How to Avoid It?

Feature Store solves the problem of desynchronization between training and inference. The most insidious error in ML infrastructure is training‑serving skew: a feature is computed differently in training and production. The model learns on correct data, but inference gets different values.

Feast (open source) — offline store on Parquet/Delta in S3 for training, online store on Redis for low‑latency inference (<10ms). Feature definitions as Python code:

from feast import FeatureView, Field
from feast.types import Float32, Int64

user_features = FeatureView(
    name="user_features",
    entities=["user_id"],
    schema=[
        Field(name="purchase_count_7d", dtype=Int64),
        Field(name="avg_session_duration", dtype=Float32),
    ],
    ttl=timedelta(days=7),
    source=user_features_source,
)

One definition, used everywhere. No discrepancies. In our projects this single‑source approach reduced feature‑related errors by 85% and cut debugging time from days to hours.

Streaming features. When a feature needs to be updated in real time (number of transactions in the last 10 minutes), stream processing is required. Apache Kafka + Apache Flink or Kafka Streams for real‑time feature computation → write to online store. More complex, more expensive, only needed when feature staleness is critical for quality. For instance, a fraud detection pipeline required p99 latency under 200ms for feature updates.

Data Labeling: How Not to Waste Budget

Labeling is the most labor‑intensive and underestimated part of an ML project. Poorly labeled data cannot be fixed by any architecture.

Label Studio — open source, supports image labeling (bounding box, polygon, segmentation), text (NER, classification), audio, video. Deploys in 10 minutes via Docker. For small teams — first choice.

Labeling quality assessment. Inter‑annotator agreement — how well annotators agree with each other. Cohen’s Kappa > 0.8 — good, 0.6‑0.8 — acceptable, < 0.6 — task ambiguous or instructions poor. Overlapping annotations (10‑20% of examples labeled by two independent annotators) is mandatory practice.

Active learning prevents budget waste. Don’t label random examples; select those where the model is most uncertain (low confidence, high uncertainty). Allows achieving the same quality with 50‑70% of the labeling volume. Modals, Prodigy, Label Studio support active learning workflows. In one NLP project, we reduced the labeling budget by 2.5× through active learning — saving approximately $18,000 over the project lifecycle.

Synthetic data. When real data is scarce or expensive to obtain. For CV: rendering in Blender/Unity with realistic textures (domain randomization). For NLP: paraphrase via LLM, backtranslation. Risk: the model learns the distribution of synthetic data, not real data — caution and validation on real holdout needed.

Data Quality: Validation and Monitoring

Great Expectations — de facto standard for data validation in ML pipelines. Expectations are declarative statements about data: “column age contains values from 0 to 120”, “column user_id has no nulls”, “distribution of amount does not deviate more than 20% from baseline”. Runs in the pipeline, on failure blocks progression. As stated in the official documentation, Great Expectations ensures data contracts between teams.

Pandera — Pythonic alternative for pandas/polars DataFrames. Schema‑based validation with type hints:

import pandera as pa

schema = pa.DataFrameSchema({
    "user_id": pa.Column(int, nullable=False),
    "score": pa.Column(float, pa.Check.between(0, 1)),
    "label": pa.Column(str, pa.Check.isin(["positive", "negative", "neutral"])),
})

Data freshness. The model expects data from the last N days. ETL fails, data is not updated — the model uses stale features. Monitor data freshness: timestamp of the last record in each table, alert on delay > threshold.

Deduplication. Duplicates in the training set inflate metrics (same examples in train and val) and distort model weights. MinHash LSH for approximate deduplication of large datasets. For exact — hash by normalized content.

Validation Tools Comparison

What Does a Data Engineering Project for ML Include?

We provide the full cycle:

• Audit of existing data and pipelines (1 week).
• Architecture design: selection of tools, formats, labeling methods.
• Implementation of ETL/ELT pipeline with validation and monitoring.
• Documentation of code and processes (model card, data card).
• Training your team on pipeline operation.
• Post‑deployment support for 3 months.
• Access to code repository and all pipeline definitions.

How We Build a Pipeline: Step by Step

1. Audit existing data. Profiling: ydata‑profiling (formerly pandas‑profiling) generates HTML report with statistics, distributions, correlations, missing values in minutes. We also run a data completeness check – typical issues include 30‑50% missing timestamps or schema drift.
2. Pipeline design. Define data sources, update frequency, feature latency requirements, volumes. Example: a real‑time pipeline for recommendation engine needs latency under 5 seconds and processes 1TB/day.
3. Implementation and testing. Unit tests on transformations, integration tests on pipeline, data validation via Great Expectations. We target 95% test coverage for transformation logic.
4. Deployment and monitoring. Alerts on freshness, quality checks, anomalies in data volumes. Typical alert threshold: no new data for 2 hours.

Storage and Formats

For most ML projects at start: Parquet in S3 + DVC for versioning. Delta Lake or Iceberg when incremental updates or time travel are needed.

Why Trust Us

We have been working in data engineering and ML for over 8 years. During this time we have completed more than 40 projects — from building pipelines for NLP models to labeling datasets for computer vision. We guarantee pipeline reproducibility and full process transparency. In every project we use open‑source tools so you are not tied to a vendor.

Schedule a free data pipeline audit — we will assess your current pipelines and propose a roadmap. Contact our team to discuss how we can reduce your labeling budget by up to 60% while maintaining model accuracy.