AI Data Orchestrator Development
Budget: $250 – $750 USD
Budget: Fixed Price — up to $500 USD. Do not apply if you cannot deliver within this budget and timeline.
Timeline: 1 week maximum.
Project:
We need a senior AI/backend engineer to build a production-ready AI-driven data retrieval agent using LangGraph or LangChain. The agent will connect directly to our Snowflake warehouse and provide a natural-language interface for business users.
This is for a fintech AI assistant platform with two surfaces:
1. Internal staff assistant
2. Customer-facing assistant
The work focuses only on the Orchestrator Agent layer. You are NOT responsible for building the chat UI, full authentication system, or API Gateway.
What You Will Build:
- AI orchestrator agent using Python with LangGraph or LangChain
- Snowflake integration for secure data retrieval
- Natural-language request handling
- Query composition and execution
- SQL and Snowpark query support
- Result formatting for business users
- Basic role-based access control
- Tenant-scoped query handling
- PII/security guardrails before query execution
- Audit logging for every pipeline run
- Error handling for failed queries, invalid requests, permission issues, and Snowflake errors
- Simple hooks so we can add more processing or external tools later
Required Workflow:
The agent should follow a clean, maintainable pipeline:
Authenticated scoped request
→ Intent parsing
→ Metric / query resolution
→ Access validation
→ Tool selection
→ Snowflake query execution
→ Result formatting
→ Audit log output
Retrieval Modes Required:
1. On-demand SQL / Snowpark query execution
2. Real-time or near-real-time update handling when Snowflake tables change
3. Scheduled extraction and delivery to:
- another Snowflake schema
- S3
- REST endpoint
Important Architecture Boundaries:
- Work starts after the API Gateway provides an authenticated, tenant-scoped request.
- Snowflake is the only database/warehouse to query directly.
- External systems such as Salesforce, HubSpot, Card Issuer APIs, FIS, JHA, or Fiserv should be represented as tool stubs only.
- Tenant isolation, PII handling, access validation, and audit logs are mandatory.
- This is a fintech project, so security-conscious implementation is required.
Preferred Tech Stack:
- Python
- LangGraph or LangChain
- Snowflake Python Connector / Snowpark
- FastAPI preferred for API layer
- Docker
- OpenAI or Claude API
- Basic JWT / tenant-scoping awareness
- LangSmith, Datadog, or simple structured logging is a plus
Deliverables:
1. Working Python codebase for the orchestrator agent
2. LangGraph or LangChain workflow with documented stages
3. Snowflake connection and query execution layer
4. SQL / Snowpark retrieval examples
5. Scheduled extraction example
6. Streaming or near-real-time update example
7. Tool stubs for external fintech/CRM systems
8. Basic RBAC and tenant validation logic
9. PII/security guardrail placeholder or basic implementation
10. Audit logging for all requests
11. Dockerfile and environment files
12. Setup guide for running locally or inside a sandbox/dev environment
13. Demo script or notebook showing all 3 retrieval modes
14. Short handoff document or Loom walkthrough explaining architecture and how to extend it
You Must Have:
- Proven experience with LangGraph, LangChain, or similar agent frameworks
- Strong Python backend experience
- Snowflake integration experience
- Experience building secure data retrieval workflows
- Ability to work fast and independently
- Good communication through text updates
In your bid, please answer:
1. Which framework will you use: LangGraph, LangChain, or another? Why?
2. Have you built AI agents connected to databases or warehouses before?
3. Have you worked with Snowflake, Snowpark, or Snowflake Python Connector?
4. How will you handle role-based access, tenant isolation, and audit logs?
5. Can you complete a production-ready version within 1 week and $500?
6. What will be your milestone plan for 7 days?
7. Can you provide daily text updates?
8. Please share 1–2 relevant examples, repos, screenshots, or short videos of similar work.
Please do not apply if you are an agency, cannot work within the $500 budget, or cannot deliver within 1 week.
Timeline: 1 week maximum.
Project:
We need a senior AI/backend engineer to build a production-ready AI-driven data retrieval agent using LangGraph or LangChain. The agent will connect directly to our Snowflake warehouse and provide a natural-language interface for business users.
This is for a fintech AI assistant platform with two surfaces:
1. Internal staff assistant
2. Customer-facing assistant
The work focuses only on the Orchestrator Agent layer. You are NOT responsible for building the chat UI, full authentication system, or API Gateway.
What You Will Build:
- AI orchestrator agent using Python with LangGraph or LangChain
- Snowflake integration for secure data retrieval
- Natural-language request handling
- Query composition and execution
- SQL and Snowpark query support
- Result formatting for business users
- Basic role-based access control
- Tenant-scoped query handling
- PII/security guardrails before query execution
- Audit logging for every pipeline run
- Error handling for failed queries, invalid requests, permission issues, and Snowflake errors
- Simple hooks so we can add more processing or external tools later
Required Workflow:
The agent should follow a clean, maintainable pipeline:
Authenticated scoped request
→ Intent parsing
→ Metric / query resolution
→ Access validation
→ Tool selection
→ Snowflake query execution
→ Result formatting
→ Audit log output
Retrieval Modes Required:
1. On-demand SQL / Snowpark query execution
2. Real-time or near-real-time update handling when Snowflake tables change
3. Scheduled extraction and delivery to:
- another Snowflake schema
- S3
- REST endpoint
Important Architecture Boundaries:
- Work starts after the API Gateway provides an authenticated, tenant-scoped request.
- Snowflake is the only database/warehouse to query directly.
- External systems such as Salesforce, HubSpot, Card Issuer APIs, FIS, JHA, or Fiserv should be represented as tool stubs only.
- Tenant isolation, PII handling, access validation, and audit logs are mandatory.
- This is a fintech project, so security-conscious implementation is required.
Preferred Tech Stack:
- Python
- LangGraph or LangChain
- Snowflake Python Connector / Snowpark
- FastAPI preferred for API layer
- Docker
- OpenAI or Claude API
- Basic JWT / tenant-scoping awareness
- LangSmith, Datadog, or simple structured logging is a plus
Deliverables:
1. Working Python codebase for the orchestrator agent
2. LangGraph or LangChain workflow with documented stages
3. Snowflake connection and query execution layer
4. SQL / Snowpark retrieval examples
5. Scheduled extraction example
6. Streaming or near-real-time update example
7. Tool stubs for external fintech/CRM systems
8. Basic RBAC and tenant validation logic
9. PII/security guardrail placeholder or basic implementation
10. Audit logging for all requests
11. Dockerfile and environment files
12. Setup guide for running locally or inside a sandbox/dev environment
13. Demo script or notebook showing all 3 retrieval modes
14. Short handoff document or Loom walkthrough explaining architecture and how to extend it
You Must Have:
- Proven experience with LangGraph, LangChain, or similar agent frameworks
- Strong Python backend experience
- Snowflake integration experience
- Experience building secure data retrieval workflows
- Ability to work fast and independently
- Good communication through text updates
In your bid, please answer:
1. Which framework will you use: LangGraph, LangChain, or another? Why?
2. Have you built AI agents connected to databases or warehouses before?
3. Have you worked with Snowflake, Snowpark, or Snowflake Python Connector?
4. How will you handle role-based access, tenant isolation, and audit logs?
5. Can you complete a production-ready version within 1 week and $500?
6. What will be your milestone plan for 7 days?
7. Can you provide daily text updates?
8. Please share 1–2 relevant examples, repos, screenshots, or short videos of similar work.
Please do not apply if you are an agency, cannot work within the $500 budget, or cannot deliver within 1 week.