Senior Fullstack AI Engineer (PwC – AI Platform, RAG)

Project

Build an AI platform (RAG, search, document drafting) to automate and scale complex corporate consolidation assessments for PwC.

• Duration: 1 year (Start: ASAP) | HackerRank: Yes
• Location: Fully remote (occasional quarterly team sessions in Prague; Prague-based candidates preferred)
• Hours: Core US overlap required (minimum 2–6 PM CET, preferred 2–7 PM CET)
• Contract: B2B only — we only accept contractors with a valid EU VAT number verifiable in VIES (VAT Information Exchange System)

Technical Stack & Core Responsibilities

1. Python Backend & Platform Engineering

• Core: Python 3.12+, OOP, clean architecture, FastAPI, CLI scripting (argparse, Bash)
• Async & Data: Celery, event-driven workflows (RabbitMQ, Azure Service Bus), PostgreSQL, MongoDB
• Data & Reporting: pandas, NumPy, openpyxl (JSON to Excel report generation), batch processing

2. RAG & LLM Engineering

• Pipeline: End-to-end RAG (ingestion → chunking → embeddings → retrieval → generation)
• LLMs & Prompts: OpenAI API (reasoning models, structured output parsing), prompt engineering and versioning (manifest.yaml)
• Search & Retrieval: Hybrid retrieval (Vector + BM25), metadata filtering, Qdrant, Azure AI Search (ACS), pgVector

3. Document Processing & Ingestion

• Extraction & OCR: PyMuPDF (fitz), Tesseract (pytesseract)
• Strategy: Clause-aware and section-aware chunking for legal/financial documents

4. Fullstack Integration & UI Demos

• Internal Tools: Multi-page Streamlit apps, session state, streamlit-aggrid for SME workflows and PoCs
• Fullstack Support: Basic-to-intermediate Node.js and React to maintain and connect frontend integration modules to Python APIs

5. DevOps & Quality Assurance

• Hygiene: Docker, Docker Compose, Git (small, regular commits), YAML-driven configurations
• Testing: Unit/integration tests for Python services and pipelines

Soft Skills & Nice-to-Haves

• Soft Skills: Autonomous pipeline debugging, strong communication to iterate on prompts/outputs alongside non-technical accounting/audit SMEs
• Nice-to-Have: RAG evaluation frameworks, MCP/FastMCP, LangChain, LangGraph, multi-agent frameworks (CrewAI, AutoGen, TaskWeave), Azure CI/CD