Description :

The Mission :

You will architect the reasoning engine of the platform. This is not about simply calling APIs. You will design autonomous AI agents that can read pitch decks, research the internet, validate information, and generate reliable, source-backed judgments without hallucinations.

Core Responsibilities :

- Build RAG Pipelines

- Parse complex PDF pitch decks

- Chunk, embed, and store data in Vector Databases

- Design retrieval pipelines that ensure accuracy and relevance

- Agent Orchestration

- Design and manage multi-agent workflows using LangChain or CrewAI

Example agents :

- Market Research Agent (uses Tavily / Serper.dev)

- Founder Background Agent (uses Proxycurl / LinkedIn APIs)

- Financial Analysis Agent

- Build coordination logic between agents

Guardrails & Accuracy :

- Enforce citation-based answers

- Implement grounding techniques to avoid hallucinations

- Validate numerical and factual outputs

- Add confidence scoring or validation layers

Python Backend Logic :

- Write clean, modular Python services

- Expose AI pipelines via APIs for the Full Stack team

- Ensure maintainability and scalability

Must-Have Tech Stack :

Language :

- Python (Expert level)

LLM Frameworks :

- LangChain

- LlamaIndex

- CrewAI

Models :

- OpenAI (GPT-4o / GPT-4.1)

- Anthropic (Claude 3.5)

- Local models via Ollama / Llama 3

Vector Databases :

- Pinecone

- Weaviate

- ChromaDB

Search & Tool APIs :

- Tavily

- Serper.dev

- Proxycurl or similar

Nice to Have :

- Experience building Agentic AI systems

- RAG optimization and evaluation techniques

- Prompt versioning and testing frameworks

- Startup / 01 product experience