This Job Description is designed for a Junior/Fresher AI Engineer specializing in the emerging Model Context Protocol (MCP) ecosystem. Given that Claude Code and MCP represent a shift toward "agentic" development, this role focuses more on orchestration, tool-use design, and terminal-based automation than traditional full-stack coding.
Job Description: Junior AI Developer (MCP & Agentic Systems)Role Overview
We are looking for a motivated Junior AI Developer to join our team in building and scaling the next generation of autonomous agents. You will be at the forefront of the Model Context Protocol (MCP), using Claude Code to develop servers that bridge the gap between Large Language Models (LLMs) and real-world data/tools.
Key Responsibilities
- Agent Development: Design and implement MCP servers using Node.js or Python to provide Claude with custom "skills" (e.g., database access, API integrations, or local filesystem control).
- Terminal-First Workflow: Use the Claude Code CLI as your primary development environment to refactor code, debug, and execute system-level tasks.
- Prompt Engineering: Optimize system prompts and tool definitions to ensure high reliability and low "hallucination" rates in agentic loops.
- Testing & Validation: Conduct rigorous testing of agentic workflows to ensure they handle edge cases and maintain security guardrails.
- Documentation: Maintain CLAUDE.md files and technical specs to ensure agents have the necessary context to remain productive across sessions.
Required Skills
- Languages: Proficiency in TypeScript/Node.js or Python.
- Environment: Strong comfort with the Terminal (Linux/macOS) and CLI-based tools.
- API Knowledge: Understanding of RESTful APIs and JSON-RPC.
- AI Familiarity: Basic understanding of LLM capabilities, specifically tool-calling (function calling) and context window management.
Fundamental Activities Every MCP Developer Must Know
To be successful, a fresher should master these five core activities. These go beyond "writing code" and enter the realm of "teaching the AI how to use tools."
1. Building an MCP Server (The "Bridge")
You must know how to wrap a standard API into an MCP server. This involves defining Resources (data the AI can read), Tools (actions the AI can take), and Prompts (templates the AI can follow).
- Task: Build a server that allows Claude to read a specific Google Sheet or local JSON database as a "Resource."
2. Mastering the "Plan-Action-Verify" Loop
In Claude Code, you don't just "run" a script. You guide an agent.
- Plan: Use /plan to let Claude outline how it will solve a bug.
- Action: Monitor the terminal as Claude edits files and runs bash commands.
- Verify: Always run tests (e.g., npm test) via the agent to ensure the fix actually works.
3. Context Management (CLAUDE.md)
Agents have a "memory" limit. A pro developer knows how to use a CLAUDE.md file in the root directory to store:
- Project-specific coding standards.
- Common build commands.
- Architectural patterns.
- Why? This prevents the agent from asking the same questions every time you start a new session.
4. Debugging via Agent Logs
When an agent fails, it’s usually because of a tool-call error or a permission block.
- Activity: Read the JSON-RPC logs between the MCP Client and Server.
- Fix: Adjust the tool definition if the AI is passing the wrong parameters (e.g., passing a string when the API expects an integer).
5. Security & Guardrails (The "Blast Radius")
Since agents can run terminal commands, you must learn to restrict them.
- Activity: Implementing read-only vs. read-write permissions for MCP servers.
- Activity: Running agents in containerized environments (like Docker) to prevent them from accidentally deleting your personal home directory.
Work Days
Monday to Saturday (6 days per week)
Total Daily Hours
8 hours per day
Core Shift Time (IST)
Approximately 5:00 PM IST to 2:00 AM IST
US Time Overlap
The core shift must overlap with USA Eastern Standard Time (EST) working hours, typically aligning from 5:00 PM IST to 1:00 AM IST.
Work Environment
Fully Remote
Collaboration Requirement
Mandatory screen-sharing connection and/or availability on call maintained throughout the full duration of the shift.
II. Initial Project Assessment
To formally start the engagement, all candidates will participate in an Initial 10-Day Project Assessment that follows the same routine and work schedule outlined in Section I. The formal employment engagement will commence only upon successful completion and review of the performance during this 10-day period.
III. Probation and Compensation StructureA. Probation Period
The initial employment period will be designated as a Probationary Period lasting between Three (3) to Four (4) Months. The exact duration will be determined by the hiring manager based on the initial assessment of the candidate's adaptation and contribution.
B. Compensation during Probation
During the entire Probationary Period, the compensation will be a fixed gross salary of ₹4,000 to ₹6,000 INR per month ( based on skill level)
C. Post-Probation Compensation (Performance-Based)
Upon successful completion of the Probationary Period, the Developer’s monthly gross compensation will be raised to a range of ₹12,000 INR to ₹15,000 INR per month.
The specific final compensation within this range will be determined based on a comprehensive performance review that assesses the following key factors:
- Skills & Technical Competence: Demonstrated technical expertise and application of best practices.
- Diligence & Reliability: Consistent work ethic, attendance, and adherence to processes.
- Delivery Quality: Quality, security, and maintainability of delivered code and features.
- Overall Contribution: Impact on project success and collaboration within the team.
IV. Collaborative Communication
The Developer is required to participate in two mandatory daily video/audio calls to ensure alignment and synchronization with the remote team structure, which overlaps with US working hours.
- Daily Morning Call: 8:00 AM IST
- Daily Evening Call: 5:30 PM IST
V. Professional Development & Reimbursement
The Company supports continuous learning through professional certification.
- Eligibility: After successfully completing the Probationary Period.
- Reimbursable Certifications: Reimbursement will be considered for official certification exams from providers such as Amazon Web Services (AWS) and Microsoft Azure.
- Condition: Reimbursement is strictly conditional upon the certification having a direct, demonstrable correlation with current or upcoming development projects within the team. The reimbursement must be pre-approved by the Lead Developer/CTO.
VI. Performance Improvement Clause (PIP)
The Company is committed to providing support to ensure consistent performance. However, a clear process is in place for addressing performance deficits.
- Initial Feedback: In the event of a noticeable dip in performance, the Developer will receive direct, constructive feedback and a performance plan within the first month of the dip being identified.
- Failure to Improve: If performance does not demonstrate adequate improvement following the initial feedback period, the Company reserves the right to implement a reduction in monthly compensation.
- Pay Reduction Basis: Any resulting reduction in pay will be proportional to the assessed impact the performance dip has on the project timelines, code quality, and team delivery capacity. This measure is intended to incentivize immediate course correction.
Pro Tip: A "Fresher" in this field is someone who stops thinking about code as a static file and starts thinking about code as a capability that an AI can exercise.
Pay: ₹4,000.00 - ₹6,000.00 per month
Benefits:
Work Location: Remote
