Autonomous Offline Python AI Development

Job ID: 39343797

Budget: $5,000 – $10,000 AUD

Title:
Build Offline Python-Based AI System with File Watcher, Reflection, and Simulation Logic

Description:
Looking for an experienced Python/AI developer to help build a modular, offline AI system that can operate fully autonomously and learn over time.

This is not a typical chatbot or cloud deployment — this system will run on a GPU-powered local environment (RTX 5080) and integrate the following components:

Core Modules Needed:

File Watcher + Ingestion

Monitor folders for new PDFs

Parse content (OCR fallback ideal)

Auto-summarize and generate threat-based Q&A

Memory System

Store extracted insights in structured files

Append to long-term memory logs

Reflection Engine

Auto-trigger summaries or reflections based on time/events

Generate insights, highlight patterns, and escalate priority topics

Simulation Framework

Run cyber scenarios with escalating difficulty

Auto-score responses and adjust future difficulty

Local Chat Interface

Use a quantized local model (DeepSeek, LLaMA, or similar)

Support config prompts, memory integration, and reflection triggers

UI (Minimal but functional)

File upload

Chat window

System status (memory, logs, GPU usage)

Must-Have Experience:

Python (multi-module, async/process mgmt)

Local LLM deployment (DeepSeek, LLaMA.cpp, or similar)

FastAPI or backend integration

GPU compute (CUDA / inference tuning)

Simulation design or gamified AI training (bonus)

Environment:

All work will run fully offline

Target device has RTX 5080, 64GB RAM

Needs to run headlessly and be Docker-deployable (optional)

Delivery Model:

Push code regularly to private GitHub repo

Milestone-based payments

You must show proof of real system logic (no screenshots-only developers)

Budget: Open depending on experience. Quality and system robustness will be prioritized over lowest cost. Looking for developers who can think ahead and build intelligently, not just implement to spec.

To Apply, Include:

Examples of similar systems or logic you've built (repos/screenshots)

Experience with offline LLMs, memory handling, or simulation engines

Availability and estimated time to deliver a working MVP