3. Data, Science & AI

Converts audio files to text using NVIDIA's Parakeet ASR model optimized for Apple's MLX framework, providing high-quality timestamped transcriptions.

Analyzes therapy transcripts and journal entries to identify cognitive patterns, emotional themes, and track progress using AI text analysis.

Data engineering tool for extracting, transforming, and analyzing NBA statistics using the nba_api library, supporting player stats, team data, and game logs.

Orchestrates Gemini and Codex AI models for deep code analysis, multi-file exploration, and multi-perspective decision-making.

Automates Kaggle competition data preparation including EDA, cleaning, and feature selection using Sweetviz, Pyjanitor, and XGBoost.

Extracts and processes YouTube video transcripts/subtitles using yt-dlp, handling auto-generated captions, rate limiting, and VTT parsing across multiple languages.

Performs psychrometric analysis for HVAC system design, calculating air properties and moisture content.

Converts Google Drive files (Docs, Sheets, etc.) by MIME type into pandas tables or local artifacts for data analysis workflows.

AI agent for HVAC fan selection and airflow analysis, delivering engineering insights for optimal system design and performance.

Provides engineering calculations and configuration guidance for HVAC air handler sizing and system design.

AI-powered agent for coil selection and thermal system design in HVAC engineering.

Optimizes database performance through query tuning, indexing strategies, and caching patterns to improve application efficiency.

Performs arithmetic, trigonometric, logarithmic, and complex mathematical calculations with precision.

Creates optimized, structured prompts for AI models based on task descriptions to improve response quality.

Enables semantic search using vector embeddings and similarity metrics including cosine, Euclidean, and dot product for effective ranking.

Provides face recognition system patterns for attendance tracking, including face detection, verification, enrollment, and liveness detection.

Extracts concise, high-signal context around ripgrep matches for optimized LLM processing, bounded by output limits.

Routes LLM requests to cost-effective models (DeepSeek, Haiku, Sonnet) for 70-90% savings without OpenAI.

Converts PDF documents to MP3 audio using MiniMax's text-to-speech AI for audiobook creation and accessibility.

Manages AI artifact tracking with YAML+Markdown format, reducing tokens by 95%. Handles creation, update, query, validation, and orchestration of AI progress artifacts.

Enables building semantic search and knowledge base systems using Weaviate vector database for efficient document retrieval.

Builds end-to-end RAG pipelines with document processing, vector search, and response generation using FastAPI, OpenAI, and Qdrant without LangChain.

Creates optimized prompts for AI coding agents and subagent delegation, enhancing task delegation and multi-step workflow efficiency in AI systems.

Enhances RAG chatbot for robotics textbook with strict grounding, citation requirements, and performance optimization.