the AI data enginethe AI data enginethe AI data engine
for modern workloads

LLMs unlocked new ways to work with data. typedef is the engine built to power them. serverless, inference-first, and built for scale.

//working session

typedef is

semantic processing, at scaleoperationalizing your AI workflowsOLAP meets LLMsbringing structure to unstructuredbuilt for text analyticsefficient rust-based computeengineering context, not just prompts

Say goodbye to brittle UDFs, hacky microservices, and fragile glue code.

The old stack wasn’t designed for inference, semantics, or LLMs.

typedef brings structure + reliability to AI-native data pipelines.

So you can build deterministic workflows on top of non-deterministic models.

Inspired by data,

optimized for AIoptimized for AIoptimized for AI

python
applicants = df.filter(
    (fc.col("yoe") > 5)
    & fc.semantic.predicate(
        "Has MCP Protocol experience? Resume: {{resume}}",
        resume=col("resume"),
    )
)

prompt = """
Is this candidate a good fit for the job?

Candidate Background: {{left_on}}
Job Requirements: {{right_on}}

Use the following criteria to make your decision:
...
"""

joined = (
    applicants.semantic.join(
        other=jobs,
        predicate=prompt,
        left_on=col("resume"),
        right_on=col("job_description"),
    )
    .order_by("application_date")
    .limit(5)
)

Semantic Operations as DataFrame Primitives

Transform unstructured and structured data using familiar DataFrame operations. If you know PySpark or Pandas, you know typedef. Semantic operations like classification work just like filter, map and aggregate.

python
df = (
    df
    .with_column("raw_blog", fc.col("blog").cast(fc.MarkdownType))
    .with_column(
        "chunks",
        fc.markdown.extract_header_chunks("raw_blog", ...)
    )
    .with_column("title", fc.json.jq("raw_blog", ...))
    .explode("chunks")
    .with_column(
        "embeddings",
        fc.semantic.embed(fc.col("chunks").content))
)

Rust-Powered Multimodal Engine

The only engine that goes beyond standard multimodal types. Native support for markdown, transcripts, embeddings, HTML, JSON and others, with specialized operations. Rust performance meets Python simplicity to process any data at any scale.

python
class Ticket(BaseModel):
    customer_tier: Literal["free", "pro", "enterprise"]
    region: Literal["us", "eu", "apac"]
    issues: List[Issue]

tickets = (df
  .with_column("extracted", fc.semantic.extract("raw_ticket", Ticket))
  .unnest("extracted")
  .filter(fc.col("region") == "apac")
  .explode("issues")
)

bugs = tickets.filter(fc.col("issues").category == "bug")

Schema driven Extraction

Type-safe structured extraction from unstructured text. Define schemas once, get validated results every time. Eliminates prompt engineering brittleness and manual validation.

python
config = fc.SessionConfig(
    app_name="my_app",
    semantic=fc.SemanticConfig(
        language_models={
            "nano": fc.OpenAILanguageModel(
                "gpt-4.1-nano", rpm=500, tpm=200_000
            ),
            "flash": fc.GoogleVertexLanguageModel(
                "gemini-2.0-flash-lite", ...
            ),
        },
        default_language_model="flash",
    ),
    cloud=fc.CloudConfig(...),
)

session = fc.Session.get_or_create(config)

Develop Locally, Scale on Cloud

Develop locally with fenic, deploy to typedef cloud instantly. Zero code changes from prototype to production. Same code, automatic scaling.

We love our

buildersbuildersbuilders

typedef is used for

conversational intelligencereal-time context engineering for AI agentsautomated content moderationlarge-scale content classificationranking and curating articlesintelligent tagging at scaleagent-based automationcustomer support automationhigh quality data labeling

“Typedef's design of Fenic is a natural evolution of the DataFrame abstraction, bringing the same clarity and composability that made pandas indispensable, now applied to modern AI and unstructured data workloads. It's exciting to see the DataFrame API extended into the AI era and love what typedef is building.”
Wes McKinney
Creator of Python pandas project, published Python author
“Typedef lets us build and deploy semantic extraction pipelines across thousands of policies and transcripts in days not months, We've dramatically reduced the time it takes to eliminate errors caused by human analysis, significantly cut costs, and lowered our Errors and Omissions (E&O) risk.”
Lee Maliniak
Chief Product Officer at Matic
“Typedef's engine gives us a powerful way to blend traditional OLAP-style analysis with LLM inference in a single, unified workflow. We're unlocked to conduct large-scale content classification for labeling, grouping, and enriching articles semantically using high-level operators, without writing brittle glue code or managing separate inference infrastructure. We're excited about the possibilities it opens up for content intelligence at scale.”
Gustavo Chavez
Ph.D., Senior Data Scientist at Medium
“LLMs are going to change how we do data science and analysis just like they’ve changed every other product category. But the breakthrough only happens when you embed them directly into the tools analysts already know and love. That’s why I’m so excited about typedef. It brings LLMs into the DataFrame itself, not as a bolt-on, but as a core primitive.”
Matt Bleifer
AI Product Manager
“Modern data platforms need to handle more than just tabular data, they need to understand free-form text, documents, and conversation. Typedef is an AI-native engine that treats inference pipelines for unstructured data as a first-class citizen. This is where data infrastructure is headed, and typedef is already there.”
Nitay Joffe
VP of Architecture at Uniphore
“Pretty excited about typedef and their new open source Fenic library. Fenic is dataframes for text data. Combined with typedef, you get AI pipelines as core primitives—like Airflow did for ETL orchestration. It’s built the way good infrastructure should be: composable, opinionated, and open.”
Chris Riccomini
Creator of SlateDB, Apache Airflow PMC, Author
“Having built data platforms at scale, I’m blown away by how typedef makes LLM inference feel like a first-class citizen in the data pipeline. It’s the first time I’ve seen unstructured AI workloads treated with the same rigor and simplicity as structured data.”
Mike Eastham
Former Founding Engineer & Chief Architect at Tecton
“Typedef transforms our OLAP warehouse into a dynamic product-signal engine by integrating LLM inference and agents. Previously, product managers spent weeks manually processing data for basic queries. Now, they can easily query and dive deep across diverse datasets, leveraging LLM categorizations and summarizations by feature, product group, or customer. This is 100x time savings and a game changer for us.”
Eric Dodds
Head of Product at RudderStack
“Typedef's design of Fenic is a natural evolution of the DataFrame abstraction, bringing the same clarity and composability that made pandas indispensable, now applied to modern AI and unstructured data workloads. It's exciting to see the DataFrame API extended into the AI era and love what typedef is building.”
Wes McKinney
Creator of Python pandas project, published Python author
“Typedef lets us build and deploy semantic extraction pipelines across thousands of policies and transcripts in days not months, We've dramatically reduced the time it takes to eliminate errors caused by human analysis, significantly cut costs, and lowered our Errors and Omissions (E&O) risk.”
Lee Maliniak
Chief Product Officer at Matic
“Typedef's engine gives us a powerful way to blend traditional OLAP-style analysis with LLM inference in a single, unified workflow. We're unlocked to conduct large-scale content classification for labeling, grouping, and enriching articles semantically using high-level operators, without writing brittle glue code or managing separate inference infrastructure. We're excited about the possibilities it opens up for content intelligence at scale.”
Gustavo Chavez
Ph.D., Senior Data Scientist at Medium
“LLMs are going to change how we do data science and analysis just like they’ve changed every other product category. But the breakthrough only happens when you embed them directly into the tools analysts already know and love. That’s why I’m so excited about typedef. It brings LLMs into the DataFrame itself, not as a bolt-on, but as a core primitive.”
Matt Bleifer
AI Product Manager
“Modern data platforms need to handle more than just tabular data, they need to understand free-form text, documents, and conversation. Typedef is an AI-native engine that treats inference pipelines for unstructured data as a first-class citizen. This is where data infrastructure is headed, and typedef is already there.”
Nitay Joffe
VP of Architecture at Uniphore
“Pretty excited about typedef and their new open source Fenic library. Fenic is dataframes for text data. Combined with typedef, you get AI pipelines as core primitives—like Airflow did for ETL orchestration. It’s built the way good infrastructure should be: composable, opinionated, and open.”
Chris Riccomini
Creator of SlateDB, Apache Airflow PMC, Author
“Having built data platforms at scale, I’m blown away by how typedef makes LLM inference feel like a first-class citizen in the data pipeline. It’s the first time I’ve seen unstructured AI workloads treated with the same rigor and simplicity as structured data.”
Mike Eastham
Former Founding Engineer & Chief Architect at Tecton
“Typedef transforms our OLAP warehouse into a dynamic product-signal engine by integrating LLM inference and agents. Previously, product managers spent weeks manually processing data for basic queries. Now, they can easily query and dive deep across diverse datasets, leveraging LLM categorizations and summarizations by feature, product group, or customer. This is 100x time savings and a game changer for us.”
Eric Dodds
Head of Product at RudderStack