On-Device Mobile AI Assistant
Budget / Salary₹250,000–500,000
TypeFreelance project
LocationRemote
Posted1 hour ago
I need a small-footprint AI that actually lives inside my phone’s internal storage and works without sending data to the cloud. Through voice or text I will say things like “open WhatsApp”, “start a new Python file”, “edit this photo”, or “type this paragraph” and the assistant should carry out the request on the spot.
Scope of control
• Mobile applications – launching, switching, closing, and basic in-app navigation
• System-level functions – toggling Wi-Fi/Bluetooth, adjusting settings, saving screenshots, etc.
• Coding and programming tasks – opening an editor, creating files, inserting sample code snippets, compiling where possible
• Typing, data entry, and general content creation or editing within any text field
Core requirements
• Runs fully on-device (Android preferred), no constant Internet dependency
• Natural-language understanding for Hindi and English commands
• Secure permission model: it must ask for any new privilege once, then remember it
• Modular design so I can add new skills later without reinstalling the whole package
• Clear documentation plus an installable APK or similar package that I can sideload
Acceptance criteria
1. Voice/text command executed within two seconds on a mid-range Android phone, offline.
2. At least 10 demo commands covering app launch, navigation, typing, and a simple code snippet insertion.
3. Source code, build instructions, and a brief video showing the assistant in action.
If you have experience with on-device NLP (e.g., TensorFlow Lite, whisper.cpp, Rasa, or similar) and Android automation APIs like Accessibility Service, I’d love to see your approach and previous work.
Scope of control
• Mobile applications – launching, switching, closing, and basic in-app navigation
• System-level functions – toggling Wi-Fi/Bluetooth, adjusting settings, saving screenshots, etc.
• Coding and programming tasks – opening an editor, creating files, inserting sample code snippets, compiling where possible
• Typing, data entry, and general content creation or editing within any text field
Core requirements
• Runs fully on-device (Android preferred), no constant Internet dependency
• Natural-language understanding for Hindi and English commands
• Secure permission model: it must ask for any new privilege once, then remember it
• Modular design so I can add new skills later without reinstalling the whole package
• Clear documentation plus an installable APK or similar package that I can sideload
Acceptance criteria
1. Voice/text command executed within two seconds on a mid-range Android phone, offline.
2. At least 10 demo commands covering app launch, navigation, typing, and a simple code snippet insertion.
3. Source code, build instructions, and a brief video showing the assistant in action.
If you have experience with on-device NLP (e.g., TensorFlow Lite, whisper.cpp, Rasa, or similar) and Android automation APIs like Accessibility Service, I’d love to see your approach and previous work.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.