Upwork home
Sign up

Development & IT Consultation with Manish P.

4.9 · 132 reviews

Development & IT Consultation with Manish P.

4.9 · 132 reviews

# AI Agent Deployment
• **Production**: Deploy on MCP (AWS, Azure, GCP) with Docker/Kubernetes. Monitor with Prometheus.
• **Voice Agents**: Use ElevenLabs (TTS), Deepgram (STT), Vapi, Retell on MCP. Link with n8n/Zapier for CRM.
• **Agent Flow**: Modular workflows via n8n/Zapier. Use MCP DBs (DynamoDB) for context.
• **Automation**: Automate with n8n/Zapier, MCP serverless (AWS Lambda). Add feedback.

# Voice Agents
• **Pipeline**: Scalable calls with ElevenLabs, Deepgram, Vapi/Retell on MCP (AWS Connect). Use n8n/Zapier.
• **Support**: AI agents with natural language, linked via n8n/Zapier.

# Prompt Engineering
• **Techniques**: Clear, role-based/chain-of-thought prompts.
• **Drafting**: Structured, clear prompts with examples.
• **Enhancing**: Few-shot learning, dynamic templates, testing.

# Business Strategy
• **Opportunities**: Automate support/sales with n8n/Zapier.
• **Integration**: Use cases (support, lead gen) with MCP, automation.
Get personalized advice on:
AI & Machine Learning AI Content Writing AI Data Annotation & Labeling AI Integration AI Speech & Audio Generation llm prompt engineering openai api gpt api voice ai voice agents

You’re covered with payment protection

Manish P.Status: Offline

About Manish

Manish P.Status: Offline
AI Agent Architect | MCP, Claude Code & Enterprise Automation
100% Job Success
4.9  (132 reviews)
Noida, India - 12:57 pm local time
🚀 Top Rated Plus | 474 jobs, 100% JSS. I build Claude Code & MCP agents that automate complex enterprise workflows in production — not just demos.

I specialize in Agentic Engineering and Multi-Agent Orchestration — designing systems where Claude Code orchestrates specialized worker agents via MCP servers to handle complex, non-linear workflows. I connect LLMs to CRM, ERP, and back-office systems, replacing manual enterprise workflows with autonomous pipelines that have monitoring, fallbacks, and cost controls built in from day one.

Context Engineering: Optimizing RAG and MCP data flows for high-precision agentic reasoning — so your agents don't hallucinate, drift, or break under load.

⚡ runloop — My open-source enterprise automation platform (MIT-licensed, clients can read the code):
• Learn-code mode: stable agent steps compile to reusable scripts — ~72% token cost reduction on recurring runs
• Multi-model routing: Claude Code, Gemini CLI, Codex — right model per step
• Human-in-the-loop: 2-minute approval windows with threaded Slack escalation
• MCP-native: unlimited MCP server connections with OAuth auto-discovery
• Computer Use: automates complex browser-based and system-level tasks with human-in-the-loop oversight
• Runs on WhatsApp, Slack, Discord, Telegram — @mention the bot to trigger workflows with approvals inline

✅ A Fit If: You're a decision-maker at a 10+ person company needing production-grade agent automation — complex multi-system orchestration, not a one-off script.

⚡ What I automate:
• Support triage (Zendesk/Intercom + CRM routing)
• Lead enrichment pipelines (HubSpot, Salesforce)
• QA regression and CI/CD agent loops
• Cloud cost optimization (AWS/GCP monitoring agents)
• Browser automation for authenticated enterprise workflows
• Custom MCP server development with OAuth auto-discovery

⚡ Recent builds:
• WhatsApp AI Recruiter — end-to-end candidate screening (DeepSeek V3)
• CanoHealth Voice AI — appointment booking for a US hospital (Mistral + NVIDIA Riva, fully open-source)
• Confida AI — legal contract review assistant

Previously Head of Engineering @ CityMall (100+ engineers, 4.5 yrs) — shipped production AI personalization that lifted revenue ~15%. Now focused full-time on enterprise agent automation.

⚡ Stack: Claude Code, Codex, Gemini CLI, MCP, n8n, Python, Golang, API Integration, RAG, vLLM

📅 Full-time — 40+ hrs/week

What to expect

Schedule the consultation
Choose from the freelancer’s available days and times.
Get advice for your custom needs
Share details about your project and what you want to talk about. The freelancer will review and reach out if they have questions.
Join the Zoom meeting
1-on-1 meeting with the freelancer to discuss your needs and project.
Approve the work
The freelancer will finish up the documents you asked for and send them to you for approval:
    Before the consultation

    Here’s what Manish will need to know before you meet

    1. Tell me about yourself and what you want to Share details about your project and what you want to talk about. The freelancer will review and reach out if they have questions. talk about
    Rating is 4.9 out of 5.
    (132)
    5 stars
    93% Complete
    (123)
    4 stars
    6% Complete
    (8)
    3 stars
    1% Complete
    (0)
    2 stars
    1% Complete
    (1)
    1 star
    1% Complete
    (0)
    GR
    Greyson R.
    5.0
    Apr 11, 2026
    MF
    Micah F.
    5.0
    Feb 20, 2026
    Manish is professional and worthwhile. I would recommend him.
    KE
    Kiran E.
    5.0
    Jul 29, 2025
    RS
    Ramya S.
    5.0
    Jun 27, 2025
    BS
    Brant S.
    5.0
    Jun 12, 2025
    Extremely knowledgeable and insightful.