Junior AI Builder

Role Overview

• Blue Sherpa builds AI systems for lenders and credit funds, helping risk teams compress the time between signal and action.
• We are hiring junior applied AI builders who like working on real problems with todays AI stack.
• This is not a traditional junior ML role, and it is not a no-code automation role.
• It is for people who like building real systems, writing code, debugging what breaks, and improving workflows until they become genuinely useful.
• The work sits at the intersection of: Applied AI and LLM systems Practical engineering Data and modeling Real business problems in credit, risk, and analytics You will work closely with experienced practitioners and take ownership of small problems end to end.
• What Youll Work On

* Build LLM-powered workflows for analytical and operational tasks

* Work with prompting, structured outputs, tool use, retrieval, embeddings, and related techniques

* Design evals to understand where a system is failing and how to improve it

* Debug issues involving model behavior, workflow logic, and messy real-world data

* Build simple end-to-end pipelines, prototypes, and internal tools

* Work on practical data and modeling problems where needed

* Turn ambiguous business problems into clear, working technical solutions

What Were Looking For

*Strong coding ability You should be able to write clean, modular, debuggable code in Python.

*Comfort with todays AI stack You should have meaningful exposure to some of the following: LLM APIs and workflows Prompting and structured generation Embeddings and retrieval Evals Fine-tuning Tool use or agent-like workflows You don't need deep expertise in everything, but you should be comfortable building with modern AI tools and understanding the tradeoffs.

Builder mindset We are looking for people who:

* Can build things outside formal assignments

* Can show real projects and code

* Can enjoy learning by doing

* Can make progress independently

* Would like figuring things out when the path is unclear Debugging instinct A strong signal is that you have built something that did not work, spent time understanding why, and improved it.

We value people who can explain what they tried, what broke, how they diagnosed it, and what they changed.

Practical problem solving You should be able to break down ambiguous problems, choose a reasonable first approach, and move from idea to working system without needing perfect instructions.

Strong Signals These are strong positive signals for us:

* GitHub or equivalent with real projects

* working AI systems you can explain clearly

* projects involving LLM workflows, evals, retrieval, embeddings, fine-tuning, or end-to-end pipelines

* evidence that you can build, not just complete coursework

* examples of independent project work

* the ability to explain what failed and how you fixed it Nice to Have

* Experience with RAG or retrieval-based workflows

* Familiarity with Hugging Face, PyTorch, or TensorFlow

* Experience with FastAPI, Flask, or similar frameworks

* Comfort with data pipelines or cloud environments

* Exposure to analytics, credit, risk, or operational use cases What This Role Is Not

* Not pure ML research

* Not a generic data analyst role

* Not just prompt engineering

* Not just no-code automation This is a builder role for people who want to work across AI, data, and systems to solve real problems.

Location-Delhi NCR,Bangalore,Chennai,Pune,Kolkata,Ahmedabad,Mumbai,Hyderabad,Remote