3. Data, Science & AI

Adds tensor shape validation and inline documentation to PyTorch models, ensuring correct dimensions and improving code clarity for machine learning development.

Handles complex-valued tensors in PyTorch for astronomical imaging, supporting Fourier transforms and phase/amplitude representations in PRISM.

Executes RAG workflows in Obsidian to answer, summarize, reorganize, or create notes with traceable citations.

Provides design guidance for OptAIC instance resources including Dataset, Signal, Experiment, Model, Portfolio Optimizer, and Backtest instances.

Manages Gemini RAG stores with Code-RAG support, enabling efficient code-based knowledge retrieval and augmentation for AI systems.

Provides semantic analysis for Obsidian vaults including concept mapping, graph generation, and embedding-based search with batch processing.

Scrapes web pages to markdown format and discovers URLs on websites using the Firecrawl tool.

Validates optical physics parameters (Fresnel numbers, diffraction regimes, resolution limits) for instruments like telescopes, microscopes, and cameras to ensure physical realism.

Provides design guidelines for Run resources in OptAIC, covering execution tracking, metrics, artifacts, and lineage for ML workflows including training, inference, and experiments.

Provides standardized MLOps patterns for model definitions, training pipelines, model registry, and monitoring in OptAIC.

Generates images from text prompts using OpenAI's GPT Image model for creative visual content creation.

Standardized workflow for manipulating XML files of any size, supporting extraction, structure modification, and validation via xpath and xmlstarlet.

Manages and queries a chronicle of world events, enabling recording, history checks, causality tracing, and event searches by type, location, or participants.

Generates images and videos using Gemini AI models, including image-to-image transformations and webcam capture via CLI commands.

Applies event-driven modifications to a simulated world state, handling entity changes from battles, treaties, and deaths.

Extracts structured data from Coles grocery invoices and predicts future orders using AI models.

Configures WaveCap to detect and prevent hallucinations in Whisper speech recognition outputs, addressing gibberish, repetition, and phantom text on silent audio.

Assesses audio transcription quality by comparing against ground truth and calculating WER/CER metrics for system evaluation.

Analyzes WaveCap transcriptions for quality, accuracy, and consistency. Compares original vs corrected text to identify issues and ensure reliable output.

Creates, validates, and queries WSC world entities stored in JSON files.

Exports WaveCap transcriptions (as JSON or with audio) and pager feed data for analysis.

Manages WaveCap configuration for Whisper speech recognition, LLM correction, and streaming parameters via config.yaml.

Run SQL queries on WaveCap SQLite database to inspect transcription records, event metadata, and historical data for analysis and debugging.

Searches WaveCap transcriptions using text queries and filters to locate specific content or historical entries.