Manish S.

Salesforce Architect with 15+ years of Salesforce experience and 20+ years inenterprise IT — delivering end-to-end implementations across Sales Cloud, Service Cloud, CPQ, Marketing Cloud, and Experience Cloud for Fortune 500 clients including Google Cloud, PG&E, CVS Health, Wells Fargo, and Credit Suisse.

As an AI Agent Engineer, I design and operate production multi-agent systems built on Claude Code, the Claude Agent SDK, and the Anthropic API, with architectures that run equally against OpenAI (GPT) and Google Gemini.

I build custom MCP servers in Python and TypeScript that connect AI agents to real business systems - CRM, billing, support desk, e-commerce - with OAuth security, typed tool schemas, and audit logging.

My systems follow a strict governance pattern: an orchestrator delegates to scoped subagents, agents review other agents’ output against written checklists, and every change passes a human approval gate before it ships.

Production deployments include AI account summaries grounded on support tickets writing into the CRM of a leading creator-economy platform, an autonomous billing follow-up agent on live invoice data at a B2B SaaS company, a scheduled e-commerce content agent generating social media campaigns from live store data, and a flat-rate AI task service I run as a product - agents doing the work, humans approving it.

I specialize in three tracks:

🔹 AI AGENTS & MCP — Production AI agent systems built on Claude (Claude Code, Agent SDK, Anthropic API), OpenAI (GPT), and Google Gemini. I build custom MCP servers (Python/TypeScript) that connect AI agents to any business system — CRM, billing, support desk, e-commerce — with OAuth security, typed tool schemas, and audit logging. My architecture pattern: an orchestrator delegating to scoped subagents, agents reviewing other agents’ output, and human approval gates before anything ships. In production today: AI account summaries grounded on support tickets for a leading creator-economy platform, a scheduled e-commerce content agent, and a flat-rate AI task service I operate as a product.

🔹 AGENTFORCE — Designing and deploying Agentforce agents, Prompt Builder templates, and Flow-invoked AI actions. Shipped a dunning-process agent in production — autonomous billing follow-up grounded on live invoice data — and authored reusable templates for case triage, service summaries, and knowledge-base Q&A.

🔹 SALESFORCE — Architecture, CPQ (SteelBrick / Revenue Cloud), Lightning (LWC + Aura), Apex, Flows, Administration, Integrations (MuleSoft, SAP, NetSuite), and DevOps (SFDX, Copado, AutoRabit,