Full Stack & AI Engineering Intern

Role Overview

Nuvama's Quant Engineering team builds internal platforms that power quantitative research and systematic trading. We are expanding our investment in AI-driven tooling and are looking for engineering interns who can contribute across the full stack — from production APIs and React interfaces to LLM integration and fine-tuning pipelines. This is a 6-month internship.

Responsibilities

• Full stack product development: building and iterating on internal research and analytics platforms using React, TypeScript, and Python (FastAPI).
• LLM integration and agent development: integrating large language models into internal tooling via API and building agent backends.
• Fine-tuning pipelines: constructing SFT datasets and running fine-tuning jobs using LoRA/QLoRA on open-source models like Llama 3, Mistral, or Phi-3.
• Retrieval-Augmented Generation (RAG): building and evaluating RAG pipelines using vector search and embedding models.
• Backend services and APIs: building REST and WebSocket APIs, async task workers, and data ingestion pipelines.
• Infrastructure and tooling: contributing to deployment automation and monitoring.

Requirements

• Pursuing B.Tech / M.Tech in CS, AI/ML, or a related field from a Tier 1 institution.
• Strong Python fundamentals and comfort with building REST APIs and async code.
• Working knowledge of React (hooks, state management) and TypeScript.
• Hands-on interest in LLMs, including prompt engineering, RAG, or fine-tuning.
• Conceptual understanding of transformer architecture (attention, tokenisation, context windows).
• Comfortable working on Linux servers and debugging distributed services.

Nice to Have

• Experience fine-tuning open-source LLMs using PEFT/LoRA or Axolotl.
• Experience building agent systems with function-calling.
• Familiarity with vector databases and RAG evaluation frameworks.
• Exposure to managed LLM inference platforms like Azure AI Foundry or AWS Bedrock.

Skills

• Python
• React.js
• FastAPI
• LLMs
• TypeScript