Custom GenAI Coding Lab Platform
Budget: ₹150,000 – ₹250,000 INR
I’m building a “learn-by-doing” lab for my Retrieval-Augmented Generation (RAG) course and need a full-stack engineer who can deliver the complete platform—from architecture to deployment.
Core layout
• Left pane: lesson instructions and guided prompts that I can edit in Markdown.
• Right pane: a live code editor with the feel of Jupyter, Colab or VS Code; it must run Python end-to-end (backend kernels, package management, plotting, etc.).
• Top-right corner: an integrated LLM assistant that offers contextual help, code generation and debugging support, always aware of the active notebook/cell.
Key requirements
• Python-only for now, but the stack can be designed so additional languages slot in later.
• Smooth inline execution—no page refresh—so students see results instantly.
• Secure sandboxing for user code.
• Session storage that persists notebooks and chat history.
• Admin interface for me to upload new lessons, monitor usage, and reset environments between cohorts.
LLM assistant
The assistant should read the current lesson, the student’s code buffer and runtime errors, then respond accordingly. I’m comfortable providing an OpenAI key, but architect it so models can be swapped (e.g., Azure OpenAI, Anthropic). Rate-limit and token-audit hooks are essential.
Expected deliverables
1. Technical blueprint: architecture diagram, data flow, and component list.
2. MVP build: responsive UI, live editor, kernel orchestration, and chat sidebar wired to my API key.
3. Deployment scripts (Docker/Kubernetes) plus CI/CD pipeline.
4. Documentation and a recorded walkthrough so future instructors can maintain the system.
5. Two rounds of post-launch refinements.
If you’ve built cloud notebooks, code sandboxes, or LLM copilots before, I’d love to see a demo link.
Core layout
• Left pane: lesson instructions and guided prompts that I can edit in Markdown.
• Right pane: a live code editor with the feel of Jupyter, Colab or VS Code; it must run Python end-to-end (backend kernels, package management, plotting, etc.).
• Top-right corner: an integrated LLM assistant that offers contextual help, code generation and debugging support, always aware of the active notebook/cell.
Key requirements
• Python-only for now, but the stack can be designed so additional languages slot in later.
• Smooth inline execution—no page refresh—so students see results instantly.
• Secure sandboxing for user code.
• Session storage that persists notebooks and chat history.
• Admin interface for me to upload new lessons, monitor usage, and reset environments between cohorts.
LLM assistant
The assistant should read the current lesson, the student’s code buffer and runtime errors, then respond accordingly. I’m comfortable providing an OpenAI key, but architect it so models can be swapped (e.g., Azure OpenAI, Anthropic). Rate-limit and token-audit hooks are essential.
Expected deliverables
1. Technical blueprint: architecture diagram, data flow, and component list.
2. MVP build: responsive UI, live editor, kernel orchestration, and chat sidebar wired to my API key.
3. Deployment scripts (Docker/Kubernetes) plus CI/CD pipeline.
4. Documentation and a recorded walkthrough so future instructors can maintain the system.
5. Two rounds of post-launch refinements.
If you’ve built cloud notebooks, code sandboxes, or LLM copilots before, I’d love to see a demo link.