Build a Multilingual Catalog Onboarding and Evidence Synthesizer — implementation

## Build
Extract product details and compliance evidence from regional supplier invoices and informal notes. Map attributes into Beckn schema catalogs with automated multilingual translations for onboarding.

## User
Catalog Operations Specialist

## Workflow
1. Ingest supplier item drafts and receipt images with regional text annotations.
2. Normalize extracted attributes into structured catalog items referencing Beckn Core schemas.
3. Invoke Bhashini pipeline mock to generate translated titles and descriptions across target languages.
4. Generate a verified Beckn catalog JSON payload ready for staging import.

## Interface
Web dashboard showing raw supplier notes, side-by-side multilingual item cards, and catalog export status.

## Data
Generate synthetic Kirana supplier invoices and product spec sheets using local fixture scripts.
Starter examples below are synthetic. They are not downloaded records.

## Source
[Beckn protocol specifications](https://raw.githubusercontent.com/beckn/protocol-specifications/master/README.md)
Implements Beckn Core discovery packet structures with context and item message catalog payloads.

## Tools
pydantic-ai, fastapi, uvicorn.
Ensures strict type validation across multi-attribute Beckn catalog payloads and nested descriptors.
Local open weights or hosted LLM with multilingual instruction-following capability.

## Deliverables
Submit runnable source, pinned dependencies, setup commands, a working interface, example outputs, and a recorded end-to-end demo. Include tests and an action log.

## Acceptance
- Correctly parses supplier attributes into Beckn descriptor, price, and category schemas.
- Translates catalog entries into at least two target Indian language codes using ISO-639 tags.
- Flags missing mandatory evidence fields including net weight and manufacturer declarations.
- Exports compliant Beckn discovery message payloads without execution errors.

## Starter Examples
```json
[
  {
    "input": {
      "supplier_id": "SUP-701",
      "source_lang": "hi",
      "raw_text": "Kalonji seed 200g, MRP 120 INR, FSSAI lic 10012000000123",
      "target_langs": [
        "ta"
      ]
    },
    "expected_output": {
      "catalog_item": {
        "id": "ITEM-SUP701-01",
        "descriptor": {
          "name": "Kalonji Seeds 200g",
          "translations": [
            {
              "code": "ta",
              "name": "கருஞ்சீரகம் 200g"
            }
          ]
        },
        "price": {
          "currency": "INR",
          "value": "120.00"
        },
        "evidence": {
          "fssai_license": "10012000000123",
          "verified": true
        }
      },
      "status": "READY_FOR_STAGING"
    }
  },
  {
    "input": {
      "supplier_id": "SUP-702",
      "source_lang": "hi",
      "raw_text": "Loose mustard oil, price 90",
      "target_langs": [
        "bn"
      ]
    },
    "expected_output": {
      "catalog_item": null,
      "status": "REJECTED_MISSING_EVIDENCE",
      "missing_fields": [
        "fssai_license",
        "net_quantity",
        "packaged_date"
      ]
    }
  }
]
```

## Safety
Uses synthetic merchant listings; enforces mock simulation of regulatory verifications without live API calls.
Use mock external writes. Require human review before any real deployment.