Contributing Writer · Retrieval & Evaluation

Soraya Zheng

Contributing Writer, ASCENDING Inc

Soraya contributes the enterprise RAG pillar: chunking, re-rankers, evaluation. She covers the practical questions platform teams face when retrieval becomes the bottleneck. Her articles anchor the RAG and evaluation sections of the glossary.

Enterprise RAG
Re-rankers
Evaluation

View LinkedIn profile

By this contributor

Articles by Soraya Zheng

22 articles published on the ASCENDING Blog

GenAI · Jarvis AI 10-19-2026 · 8 min read

Moving from Codex to Claude Code: a Team Assessment Checklist

Native import can copy configuration, not prove a safe team cutover. Map Codex workflows, test Claude Code controls and record the gaps before switching.
GenAI · Jarvis AI 10-19-2026 · 7 min read

MCP Audit Logging: Evidence to Collect and Tests to Run

A tool-call dashboard isn't an audit trail. Map identity, authorization and outcomes across MCP systems, then test the evidence without storing secrets.
GenAI · Jarvis AI 10-19-2026 · 7 min read

MCP Tool Poisoning: Review Changes Before Reapproving a Tool

An approved MCP server can expose a changed tool definition later. Use a review record and safe tests to decide whether that update should reach company users.
GenAI · Jarvis AI 10-16-2026 · 7 min read

Moving a Company from ChatGPT to Claude: What to Assess

A company AI migration is more than moving chats. Map workspace assets, API dependencies and permissions, then test one workflow before committing to a rollout.
Jarvis AI · GenAI 10-01-2026 · 6 min read

AI Skills vs. MCP Tools: What Should Your Team Govern?

Compare AI skills and MCP tools, then see how Jarvis governs private skill distribution and live tool access without treating one as the other.
Jarvis AI 10-01-2026 · 6 min read

Should We Build Our Own AI Agent Infrastructure or Buy It?

Compare an internal AI agent build with Jarvis orchestration and governance. Review supplied capabilities, cloud delivery scope and cost per completed task.
Jarvis AI · Data Analytics 10-01-2026 · 5 min read

Can Our Team Ask Questions About Business Data Without Writing SQL?

See how Jarvis AI supports plain-language warehouse questions, with a customer example and practical checks for data access, metrics and answer accuracy.
Jarvis AI 10-01-2026 · 5 min read

Can We Run a Company Knowledge Assistant in Our Own Cloud?

See how Jarvis Chat and Knowledge Base support a company knowledge assistant in your cloud, with source citations, employee access and deployment checks.
Jarvis AI · GenAI 10-01-2026 · 6 min read

Customer-Hosted Glean Alternatives: What to Compare

Compare Jarvis, Onyx and Glean for customer-hosted company knowledge: operating responsibilities, source permissions, model data paths and actions.
GenAI · Jarvis AI 10-01-2026 · 5 min read

LLM Fine-Tuning Services: What Should You Receive?

Compare LLM fine-tuning services by data preparation, evaluation, model artifacts, deployment, and handoff—not just the training job or model benchmark.
Jarvis AI · GenAI 10-01-2026 · 6 min read

How to Control MCP Tool Access with Microsoft Entra ID

Evaluate MCP access control with Microsoft Entra ID: how Jarvis connects sign-in, roles, resource permissions and tool-call checks in one rollout.
GenAI · Jarvis AI 10-01-2026 · 5 min read

Switch AI Models Without Rebuilding Apps: What to Check

Evaluate AI gateways for switching an application's models. Compare API compatibility, provider routes and testing, and when a multi-model chat workspace fits.
Jarvis AI · GenAI 09-28-2026 · 7 min read

AI Insurance Eligibility Verification: A Workflow Guide

Design AI insurance eligibility verification in 4 steps: intake, payer evidence, exception routing, and reviewed updates. Scope a governed Jarvis workflow.
Jarvis AI · GenAI 09-28-2026 · 8 min read

Automated Insurance Verification: Pros, Cons and ROI

Weigh automated insurance verification pros and cons with a 5-step pilot and an illustrative ROI example. Compare staff effort, exceptions, and full costs.
GenAI · Jarvis AI 09-23-2026 · 5 min read

Private LLM Deployment: What Buyers Should Verify

Compare private LLM deployment options before choosing a partner. Check where inference runs, who controls data, and who owns security and operations.
GenAI · Jarvis AI 09-23-2026 · 5 min read

RAG vs. Fine-Tuning for a Private LLM

Choose RAG, fine-tuning, or both for a private LLM project. Compare data needs, source attribution, evaluation, and the cost of maintaining each approach.
GenAI · Jarvis AI 09-15-2026 · 5 min read

Can AI Automate Business Workflows and Still Require Human Approval?

See how Jarvis AgentFlow combines workflow automation with human approval. Follow a purchase request and check reviewer, permission and retry behavior.
GenAI · Jarvis AI 09-15-2026 · 4 min read

AI Token Cost Tracking: Users, Teams and Budgets

Evaluate AI token cost tracking by user, team, model, and application. Separate billing, traces, budget alerts, and enforceable spending controls.
GenAI · Jarvis AI 09-15-2026 · 4 min read

Amazon Bedrock Migration Assessment: A Buyer's Checklist

Assess an OpenAI or Anthropic application before moving to Amazon Bedrock: model fit, data boundaries, cost, testing, rollout, and operational handoff.
GenAI · Jarvis AI 09-15-2026 · 5 min read

Can We Add an AI Assistant to Our Employee Portal with Company Sign-In?

Add an AI assistant to an employee portal with Jarvis Chat. Review embedding, company sign-in, source-backed answers and a practical IT-help pilot.
GenAI · Jarvis AI 09-15-2026 · 5 min read

Enterprise RAG: Metadata, Hybrid Search and Embeddings

Choose how enterprise RAG retrieves documents, metadata and database records. Compare embeddings, access controls and tests before selecting a platform.
Jarvis AI · GenAI 09-03-2026 · 8 min read

Enterprise MCP Registry Architecture: Identity, Policy & Audit

A practical enterprise MCP registry architecture for governing discovery, identity, per-tool policy, lifecycle, runtime calls, and audit across AI clients.

Our contributors

Soraya Zheng

Articles by Soraya Zheng

Moving from Codex to Claude Code: a Team Assessment Checklist

MCP Audit Logging: Evidence to Collect and Tests to Run

MCP Tool Poisoning: Review Changes Before Reapproving a Tool

Moving a Company from ChatGPT to Claude: What to Assess

AI Skills vs. MCP Tools: What Should Your Team Govern?

Should We Build Our Own AI Agent Infrastructure or Buy It?

Can Our Team Ask Questions About Business Data Without Writing SQL?

Can We Run a Company Knowledge Assistant in Our Own Cloud?

Customer-Hosted Glean Alternatives: What to Compare

LLM Fine-Tuning Services: What Should You Receive?

How to Control MCP Tool Access with Microsoft Entra ID

Switch AI Models Without Rebuilding Apps: What to Check

AI Insurance Eligibility Verification: A Workflow Guide

Automated Insurance Verification: Pros, Cons and ROI

Private LLM Deployment: What Buyers Should Verify

RAG vs. Fine-Tuning for a Private LLM

Can AI Automate Business Workflows and Still Require Human Approval?

AI Token Cost Tracking: Users, Teams and Budgets

Amazon Bedrock Migration Assessment: A Buyer's Checklist

Can We Add an AI Assistant to Our Employee Portal with Company Sign-In?

Enterprise RAG: Metadata, Hybrid Search and Embeddings

Enterprise MCP Registry Architecture: Identity, Policy & Audit

Other authors