Artificial Intelligence Engineer Intern

Company Description

InfraDock AI is an early-stage technology startup focused on bringing cutting-edge artificial intelligence to the construction industry.

Our team is developing a platform that will use advanced AI to understand, analyze, and extract insights from construction documents at scale. By streamlining document-heavy workflows and bringing critical insights to the surface, we enable construction teams to focus on execution and quality.

Role Overview:

AI Engineer

1. Designing custom domain-aware ingestion processes to interpret project text documents like a domain-expert
2. Researching and implementing cutting-edge RAG techniques to enhance retrieval and reasoning systems
3. Building a professional-grade Agentic Workflow to bring intelligence and insights from unstructured project documents directly to industry experts

This role is hands-on — you’ll design, build, and optimize LLM pipelines, embedding models, and chunking logic to improve retrieval and reasoning quality across large, heterogeneous EPC documents.

You’ll Be a Great Fit If You Have:

Prerequisites:

• B.Tech / M.Tech / B.Sc / M.Sc in CS, AI/ML, Data Science, or related fields
• Work or project experience in building smart RAG pipelines over unstructured data
• Hands-on with prompt engineering and fine-tuning AI agents
• Working knowledge of LangChain / LangGraph/ CrewAI and Agentic Workflows

Technical Skills Required:

• Python expert: NumPy, pandas, dataclasses, type hints
• Strong fundamentals in Natural Language Processing (NLP)
• Experience with and strong understanding of RAG architectures: Chunking strategies, metadata enrichment etc
• Hands-on with vector databases: FAISS / ChromaDB / Pinecone / Elasticsearch
• Project experience in using document parsing tools (Unstructured.io, PyMuPDF, PDFPlumber, OCR) for multi-format unstructured files
• Solid understanding of different embedding models (OpenAI/Instructor/MiniLM/Mistral embeddings)
• Solid understanding of information retrieval techniques: Similarity search, neighbor expansion, context-aware retrieval, contextual compression etc
• Experience with different Reranking methods and models like BGE/MS-Marco Cross-Encoder Rerankers

What You’ll Build / Learn:

During the Internship:

• Understand and then build on the current RAG + LLM architecture
• Enhance and test chunking, embedding and retrieval logic
• Implement cross-encoder reranking and query decomposition models/agents
• Optimize retrieval scoring, clustering, and reasoning chains
• Work on end-to-end multi-agent workflows and integrate them with orchestration frameworks like LangGraph / LangChain / CrewAI etc

By the End of the Internship:

You’ll have deep, hands-on experience in:

• Building production-grade RAG + multi-agent pipelines
• Designing domain-aware AI systems for multi-format document reasoning
• Working with LangGraph / LangChain
• Real-world application of LLM reasoning, retrieval, and orchestration

Soft Skills:

• Strong analytical and research-based mindset, with a drive to solve real-world industry level problems
• Enjoy taking on responsibilty, brainstorm ideas and experiment with new methods
• Strong interest in enhancing retrieval techniques,and constructing agentic workflows
• Comfortable reading research papers / articles and translating into code

Duration:

Location

Why Join Us:

• Work on real AI systems, not toy projects
• Build cutting-edge AI solutions for industry-specific problems
• Build your portfolio with production-grade RAG and agentic workflows
• Potential to convert to full-time after internship