Building reliability infrastructure for AI agents
Founder of Vouqis, a reliability gateway for MCP servers.
Focused on AI agent reliability, MCP infrastructure, observability, and protocol-level failures.
Current Focus: MCP Reliability · Agent Infrastructure · Observability · AI Systems
Vouqis is a reliability gateway for MCP servers.
AI agents often receive successful responses even when tool outputs are empty, malformed, incomplete, or unusable.
Traditional monitoring sees:
HTTP 200 OK
Users experience:
Task failed
Vouqis sits between the agent and MCP server and catches these failures before they reach production.
Agent
↓
Vouqis Gateway
↓
MCP Server
Current capabilities:
- Request validation
- Response validation
- Timeout detection
- Retry logic
- Structured audit logs
The goal is simple:
Detect failures that appear successful.
Repository:
https://github.com/Sasisundar2211/Vouqis
While building AI agents and automation systems, I repeatedly encountered failures that looked successful.
The tool call succeeded.
Logs showed success.
The agent continued execution.
The user still received a bad result.
The hardest failures were not crashes.
They were failures hidden behind successful responses.
After seeing this pattern repeatedly, I started building tools focused on reliability instead of capability.
Most AI infrastructure focuses on:
- Better models
- Better prompts
- Better orchestration
A large percentage of production failures happen somewhere else:
- Tool calls
- Protocol boundaries
- Invalid responses
- State mismatches
- Silent failures
Reliability becomes the bottleneck before intelligence does.
Reliability gateway for MCP servers.
Focus:
- Response validation
- Failure detection
- Reliability enforcement
- Production observability
Repository:
https://github.com/Vou-qis/vouqis
Multi-agent system designed to detect procurement inconsistencies and support contract compliance workflows.
Highlights:
- Structured document extraction
- Workflow automation
- Agent orchestration
- Reliability-focused execution
Tech:
Python · FastAPI · Gemini · Docker
Repository:
https://github.com/Sasisundar2211/Autonomous-Procurement-AI-System
Experimental environment for studying tool orchestration, retries, execution flow, and agent reliability.
Focus:
- Tool chaining
- Deterministic execution
- Agent evaluation
- Reliability testing
Tech:
FastAPI · Google ADK · LangChain
Repository:
https://github.com/Sasisundar2211/sample_multitool_agent
Workflow automation experiments using n8n.
Focus:
- Process automation
- Operational efficiency
- Workflow orchestration
Repository:
https://github.com/Sasisundar2211/n8n_Workflows
- AI Agents
- MCP Infrastructure
- Agent Reliability
- Tool Orchestration
- RAG Systems
- Evaluation Systems
- Python
- FastAPI
- TypeScript
- API Systems
- Reliability Engineering
- Docker
- CI/CD
- Observability
- Cloud Deployment
- Production Monitoring
- Reliability over features
- Evidence over assumptions
- Production reality over local success
- Customer problems over technical novelty
- Deterministic systems over hidden behavior
- Measure outcomes, not activity
Understand why AI agents fail in production and build infrastructure that prevents those failures before they reach users.
- Founder, Vouqis
- Final-year B.Tech AIML student
- Internship at BITS Pilani Hyderabad
- Microsoft & SAP TechSaksham participant
- Co-authored applied IoT research paper
- Built multiple AI systems across automation, agents, and infrastructure
LinkedIn: https://www.linkedin.com/in/sasi-sundar
Email: sasisundhar2211@gmail.com