3. Data, Science & AI

Guides end-to-end development of LLM-driven projects, including agent design, approach selection, and batch processing pipelines.

Optimizes long-running AI agent sessions by compressing conversation history to reduce memory usage and token consumption without losing critical context.

Creates AI agent workflows enabling tool usage, conversational capabilities, and multi-step reasoning for complex task automation.

AI-powered agent enabling natural language crypto trading, portfolio management, and DeFi operations across multiple blockchains.

Validates AI implementation plans through parallel analysis and devil's advocate review to ensure robustness before execution.

Validates dbt models and schema YAML files for data quality, detecting anti-patterns and suggesting fixes during migrations.

Specializes in database schema design, query optimization, and data modeling to enhance database performance and structure.

Creates AI agent skills by authoring system prompts and implementing Chain of Thought and Few-Shot patterns for optimized agent behavior.

Coordinates multiple AI agents in parallel for complex task decomposition and execution within OpenCode swarm workflows.

Provides reusable patterns for building efficient data synchronization with Nango, including pagination, batch processing, and incremental updates.

Enables seamless integration with Orama for search, indexing, and answer engine functionality in data-driven applications.

Manages Obsidian BASE databases with filters, properties, and dynamic views for structured data organization.

Creates and manages Obsidian Bases for database-like organization of notes, supporting views, filters, and formulas to structure personal knowledge.

Writes production-ready SurrealDB queries and operations using SurrealQL for schema creation, CRUD, graph relationships, and authentication.

Generates consensus answers from multiple AI models (Claude, Codex, Gemini) for complex problem-solving.

Provides model interpretability and explainability for machine learning models using SHAP, including feature importance and various plot types for analysis.

Queries Reactome REST API for systems biology analysis, including pathway mapping, gene interactions, and disease pathway exploration.

Infers transcription factor-target relationships from scRNA-seq data using GRNBoost2 and GENIE3 with Dask scalability.

Detects system resources and provides strategic recommendations for computational approaches in scientific tasks.

Provides fast nonlinear dimensionality reduction using UMAP for 2D/3D visualization and clustering preprocessing of high-dimensional data.

Builds and deploys bioinformatics pipelines using Latch SDK with Nextflow/Snakemake integration for data processing workflows.

Enables fine-tuning, inference, and training of transformer-based models for NLP, computer vision, and multimodal AI tasks using HuggingFace.

Query and analyze the CZ CELLxGENE Census database for single-cell expression data with cell type, tissue, and disease filtering, integrating with scanpy and PyTorch.

Automates access to Benchling R&D platform data including biological registries, inventory, ELN entries, and workflows via API for lab data management.