About the role
We are looking for someone who has moved past "I use ChatGPT" and into "I build things with LLMs."
At Nexus Growth Advisors, AI is not a side experiment — it is becoming part of what we deliver. You will work on two fronts. Internally, you will automate the repetitive work that eats our teams' days: research, documentation, reporting, data extraction, follow-ups. Externally, you will help design and build AI solutions for our clients — custom agents, document-processing pipelines, internal copilots, workflow automation — and sit in the room while we scope them.
That second half is what makes this role different from most AI internships. You will talk to clients. You will hear a vague business problem, translate it into something an LLM can actually do reliably, and then build it. You will also be the person who says "this is not a good use of AI" when that is the honest answer.
We are hiring for capability, not credentials. If you have built a working agent — even a scrappy one — and can walk us through why you made the architectural choices you did, your degree does not matter to us.
What you'll do
Building
- Design and build custom AI agents — tool-calling, multi-step workflows, agents that take real actions in real systems
- Build RAG pipelines over document sets: chunking, embeddings, vector stores, retrieval quality, evaluation
- Write prompts as engineering artifacts — versioned, tested, with clear failure modes documented
- Integrate LLM APIs (OpenAI, Anthropic, Google, open-weight models) into working applications
- Connect AI systems to the tools a business actually runs on, via APIs, webhooks and MCP servers
- Build automations using no-code / low-code platforms (n8n, Make, Zapier, Power Automate, Apps Script) where that's the faster answer
- Build dashboards and internal tools so people can see and use what you've built
Client-facing
- Join client discovery calls, understand the business problem underneath the request, and shape it into a scoped AI solution
- Build demos and proofs of concept fast enough to be useful in a sales conversation
- Deploy solutions at client sites, train their teams, and support them after go-live
- Write documentation and SOPs a non-technical client can actually follow
Judgement
- Evaluate and stress-test AI outputs before anything reaches a client — accuracy, hallucination, edge cases
- Track what's genuinely new in the LLM and agent space and give us a straight read on what's worth adopting
- Handle client data with strict confidentiality; understand where data goes when you call an API
What we're looking for
Must have
- Hands-on experience building with LLMs beyond chat — API calls, prompt engineering, structured outputs
- Working understanding of how agents function: tool use, function calling, memory, planning loops, why they fail
- Understanding of RAG — embeddings, vector databases, retrieval, and why retrieval quality is usually the bottleneck
- Python (or JavaScript) at a level where you can build and debug a working application
- Comfort with APIs, JSON, and reading technical documentation without hand-holding
- Clear spoken and written English — you will be in front of clients
- Ability to explain a technical decision to someone with no technical background
- Willingness to work in-office in Delhi
Strongly preferred
- Any agent or LLM framework: LangChain, LangGraph, CrewAI, LlamaIndex, Claude Agent SDK, OpenAI Agents SDK, or your own from scratch
- Vector databases: Pinecone, Weaviate, Qdrant, Chroma, pgvector
- Familiarity with MCP (Model Context Protocol) or similar tool-integration standards
- Experience evaluating LLM output systematically rather than by vibes
- Any exposure to finance, accounting, tax, consulting or business operations
- Deployment experience — getting something off your laptop and into someone else's hands
Pay: ₹10,000.00 - ₹25,000.00 per month
Benefits:
- Paid sick time
- Paid time off
Work Location: In person
