How amiralishar/buildables-ai-internship works
Project Type and Purpose
• This repository is a multi‐week AI internship “project” made up of discrete weekly modules (week01, week2, …).
• Each weekly module is essentially a small full-stack/text-processing pipeline:
• a frontend (CLI or script)
• a “main” orchestration entry point
• a data directory for sample inputs and outputs
• utility libraries for schema definition, tokenization, LLM prompting
• configuration, tests, and documentation
Top-Level Components
• week01/ and week2/: two separate but similarly structured sub-projects
• README.md: high-level description (currently minimal)
• Each week folder contains:
• config.py (configuration parameters, e.g., API keys, paths)
• data/ (raw sample texts and output/result folders)
• utils/ (Schema.py, llm_helpers.py, tokenizer_helpers.py or chatbot implementations)
• main.py (orchestrates data ingestion → processing → write outputs)
• frontend.py (user-facing script, possibly CLI)
• docs/ or test/ for usage examples and unit tests
• requirements.txt
Architectural Patterns
• Lightweight ETL/processing pipeline pattern
• Separation of concerns:
• “frontend” layer for user interaction
• “application” layer (main.py) for orchestration
• “service” layer (utils/*) for reusable business logic (LLM helpers, tokenizers)
• “data” layer (data/*) for input/output storage
• “config” layer (config.py) for externalized parameters
• No external microservices— everything runs locally as scripts
a. Main Components
1. Frontend Script
– week01/frontend.py, week2/frontend.py
– CLI or entry-point that triggers the pipeline
2. Orchestrator
– week01/main.py, week2/main.py
– Coordinates: read config, load inputs, call utils, save results
3. Configuration
– week01/config.py
– Central place for API keys, file paths, prompt templates
4. Data Storage
– week01/data/…: raw sample texts & results
– week2/ “Personal Response” folder for outputs
5. Utilities / Services
– Schema.py: Defines input/output data models
– llm_helpers.py: Wraps calls to the language model (e.g., OpenAI API)
– tokenizer_helpers.py: Token limit management, chunking
– Chatbot implementations in week2/utils
6. Tests & Docs
– week01/test for unit tests
– week01/docs for usage examples
b. Relationships & Interactions
– Frontend → Orchestrator: user invokes frontend, which calls main
– Orchestrator → Config: reads parameters
– Orchestrator → Data: loads raw text files
– Orchestrator → Utilities: passes text + schema to llm_helpers → receives structured output
– Utilities → External LLM Service: API calls to OpenAI (or similar)
– Orchestrator → Data: writes output files under results/ or Personal Response/
– Tests & Docs are external, used by developers only
c. Architectural Patterns / Principles
– Single Responsibility: each module has one clear role
– Modular Utility Services: Schema, LLM wrappers, tokenizer are reusable
– Layered Design: Presentation (frontend), Application (main), Service (utils), Data
d. Key Technologies & Frameworks
– Python 3.x
– OpenAI or similar LLM API (implied by llm_helpers)
– Pytest for testing (test_main.py)
– CLI argument parsing (if used in frontend.py)
– JSON / text file I/O