3. Data, Science & AI

Guides solving ARC-AGI pattern recognition tasks using git operations to merge branches and implement algorithmic transformations based on input-output examples.

Guides querying ML model leaderboards (MTEB, HuggingFace) for top-performing models, performance comparison, and live benchmark data access while avoiding outdated sources.

Modernizes legacy Python 2 scientific code to Python 3 using NumPy, pandas, and scipy for data processing and analysis pipelines.

Guides frame-level video analysis, event detection, and motion pattern extraction using OpenCV and computer vision techniques.

Guides implementation of PyTorch pipeline parallelism for distributed training of large language models, covering model partitioning and communication patterns.

Specialized tool for designing primers to insert DNA sequences into plasmids using Q5 site-directed mutagenesis, including Tm calculation and verification.

Optimizes MuJoCo MJCF model files for simulation performance while preserving numerical accuracy through physics parameter tuning and configuration adjustments.

Extracts weight matrices from black-box ReLU networks via input-output queries for model extraction and reverse engineering.

Optimizes LLM inference request batching and scheduling to minimize cost while meeting latency and padding constraints.

Guides semantic document retrieval using text embeddings, including encoding, cosine similarity computation, and ranking for MTEB benchmark and search applications.

Guides SQL query optimization by rewriting slow queries for performance while maintaining semantic equivalence, focusing on complex joins, correlated subqueries, and CTEs.

Guides optimization of MuJoCo MJCF files for improved performance without compromising physics accuracy, balancing speed and precision in simulations.

Analyzes log files across date ranges, counts entries by severity, and generates summary statistics and CSV reports.

Guides implementation of PyTorch pipeline parallelism for distributed model training, including model partitioning and inter-rank tensor communication.

Analyzes chess board images to determine optimal moves, checkmate, or tactical solutions using image-based analysis and move verification.

Guides Bayesian Network tasks including structure learning, parameter fitting, causal interventions, and sampling using pgmpy or similar libraries.

Guides minimal-code implementation of GPT-2 neural network inference, including checkpoint parsing and transformer architecture in low-level languages under strict size constraints.

Guides building Caffe from source, preparing CIFAR-10 dataset, and training CNN models with Caffe solvers on CPU.

Provides a guide for implementing reversible data resharding systems with hierarchical constraints for dataset reorganization and full reconstruction.

Guides recovery of PyTorch model architectures from state dictionaries, fine-tuning specific layers, and converting models to TorchScript format for deployment.

Guides cell segmentation using MobileSAM/SAM, including mask refinement, polygon extraction, and integration into image processing pipelines.

Provides guidance for writing and verifying SPARQL queries on academic RDF datasets, including complex graph patterns, filters, and aggregations for university ontologies.

Configures LLM models for Letta agents via SDK/API, adjusting parameters like temperature, tokens, and provider-specific settings including reasoning.

Guides recovery of SQLite database data from corrupted, encrypted, or inaccessible Write-Ahead Log (WAL) files.