3. Data, Science & AI

Executes AI model optimization loops, generating standardized cloud prompts for parallel evaluation and comparison.

Offers strategic guidance for AI agent development, including evaluation methods, dataset requirements, and improvement strategies for building and iterating on AI agents.

Expert in PostgreSQL schema design, performance tuning, and advanced features including JSONB, partitioning, and window functions.

Enables seamless integration of Langdock API for AI-powered features in applications, including client-side calls and backend services.

Executes Langfuse experiments, analyzes failures, compares evaluation cycles, and documents findings for LLM application evaluation.

Analyzes user metrics and data to identify trends, extract insights, and provide actionable feedback for performance review.

Provides real-time performance metrics overview for Ada AI agents, including CSAT, answer rate, and conversation volume for quick status updates.

Specializes in SQLite database design, optimization, and best practices including WAL mode, FTS5, and PRAGMA configuration.

Analyzes Ada AI agent data to optimize automated resolution rates, providing actionable recommendations for reducing handoffs and improving conversation resolution.

Simulates Ada AI agent conversations to test and validate responses before live deployment, ensuring configuration changes function correctly.

Audits Ada AI agent configurations including playbooks, coaching, actions, and custom instructions for user setup inquiries.

Creates and configures Claude Code subagents for task-specific workflows, enabling tailored AI agent deployment and management.

Validates Langfuse prompt schemas against output contracts to ensure consistency in LLM application development.

Enables unified access to 400+ LLM models from 70+ providers via OpenRouter API for chat, streaming, and model routing in applications.

Sets up Langfuse instrumentation for tracing LLM calls, enabling observability and scoring in AI pipelines.

Diagnoses performance issues and root causes in Ada AI agent metrics, including CSAT and AR drops, to identify underlying problems.

Integrates ZhipuAI's GLM models and multimodal AI capabilities for chat, vision, image, video, and audio tasks in applications.

Proactively resets context window at 70% capacity for AI agents to prevent overflow during extended tasks.

Automates memory management in OpenMemory Local, including connectivity verification, search, and storage pattern configuration for AI applications.

Designs evaluation metrics, dataset selection, and annotation strategies for AI agent optimization with structured workflows.

Retrieves Salesforce records (Accounts, Opportunities, Contacts, Leads, Cases) via natural language queries.

Integrates Rapier physics engine with React 3 Fiber (R3F) for game development, enabling physics simulation, colliders, and rigid bodies.

Generates expertly-formulated system prompts for Langdock assistants with German language support and tool integration specifications.

Analyzes multi-turn conversation sessions, debugs issues, compares flows, and provides session-level metrics for AI chat applications.