Honestly, the first time most of us built something with an LLM, it worked beautifully once & then fell apart the second time we ran it. That second run is where the real engineering starts. It’s where we spend our days. At SynaptyX, we build AI agents that work in the real world. Less hype, more shipping. If you like figuring out how things work, and you’d rather learn by building than by watching tutorials, you’ll probably be unreasonably happy here. The Role You’ll join the team behind our agentic orchestration platform, which brings together accelerators across AI Strategy, Data Intelligence and Analytics. You’ll work on real client facing accelerators , starting with small, well-scoped tasks and taking on more as you show what you can do. Your work gets reviewed, explained and shipped. We describe ourselves as Human, Stoic and Nerdy. Day to day, that looks like this:
• A figure-out-able mindset. Most problems look impossible until you read the docs, break them down and try something. We expect you to have a proper go first, then ask well.
• Evidence over opinion. If an agent’s answer matters, we evaluate it.
• Ten-day sprints. Every sprint starts with a scoped plan on Day 1 and ends with working software, deployed and demoed, on Day 10. Daily check-ins keep us honest about what is actually done.
• Ship, then improve. Small, real releases beat big plans. What You’ll Work On
• Building and improving AI agent workflows in Python, using LangChain and LangGraph
• Building agent tools and skills, and connecting agents to data and external systems
• Building and integrating MCP servers and clients
• Writing tests and evals to check that AI answers are accurate and reliable
• Writing backend APIs built with FastAPI
• Using AI-assisted coding tools every day, and doing it properly




• Learning how AI products are deployed and monitored Required Skills
• Solid Python and FastAPI fundamentals: you can read unfamiliar code, debug it and structure your own
• Hands-on project work with LangChain or LangGraph (coursework, personal projects or internships all count, as long as you can explain what you built)
• A clear grasp of how an LLM application differs from a single prompt: tools, state, memory, retries
• Working knowledge of SQL and REST APIs
• Comfort with Git and the command line
• Regular use of AI coding tools (Claude Code, Cursor, Copilot or similar), used with discipline. For us that means: you read and understand every change before it goes into a commit you give the tool small, well-scoped tasks with enough context to do them you write or run tests to check what it produced you never paste secrets or client data into a tool you can explain any line you submit, whoever or whatever wrote it Willingness to read documentation and source code. Bonus Points (very real ones)
• You’ve built an agent, a RAG pipeline or an MCP server, even a small one, and can tell us where it broke
• Docker and any cloud experience (AWS, Azure, etc.)
• Experience with evals, LLM-as-judge, or measuring answer quality
• Familiarity with deep agent skills — OpenSearch, Redis, Postgres, etc. Don’t Apply If
• You’ve only run notebooks and never built anything others could use
• You’d rather copy-paste an error than read it
• You accept AI-generated code without reading it
• You want a research role with no engineering in it What You’ll Get
• Real production work, not a side project
• A small, supportive team that explains the why
• Clear mentorship and regular feedback
• A path to a full time role based on what you ship

📌 AI Apprentice (India)
🏢 Synaptyx AI
📍 India