AI/ML Developer

Vakeel AI Solution Private Limited; New Delhi; On-site; INR 1,800,000 - 3,000,000; Full time

Job summary

VakeelAI in India is seeking a hands-on AI/ML engineer to design and optimize retrieval-augmented generation systems, focusing on LLM orchestration, vector search, and GraphRAG. You will work on OCR and document understanding, improving search relevance and reducing hallucinations while collaborating with backend teams to deploy APIs.

You will design and ship production-grade RAG strategies, evaluate LLM providers, and ensure accuracy and grounding of outputs in a compliance-sensitive domain.

Qualifications

• Proficient in Python with hands-on experience in LLM orchestration frameworks (e.g. LangChain).
• Strong experience with vector databases and semantic search.
• Design and ship production-grade RAG systems; GraphRAG knowledge is a plus.
• Familiar with NLP embedding libraries and model deployment workflows.
• Exposure to OCR/document parsing pipelines is beneficial.

Responsibilities

• Design and improve retrieval-augmented generation (RAG) pipelines for legal-domain data.
• Build and fine-tune embedding and retrieval strategies for domain-specific text.
• Work on OCR and document-understanding pipelines for scanned documents and PDFs.
• Improve search relevance and reduce hallucination in AI outputs.
• Collaborate with backend engineers to deploy models as APIs.
• Evaluate and compare large language model providers (hosted and self-hosted).
• Own accuracy and grounding of AI outputs, flagging low-confidence results.

Skills

Python

LangChain

Vector databases

RAG systems

NLP embedding models

OCR/document parsing

Job description

About VakeelAI: VakeelAI is an Indian legal-tech platform building AI-powered tools for lawyers, spanning legal research, search, drafting, and case management. We're solving hard retrieval, reasoning, and document-understanding problems on large, real-world text corpora.

What You'll Do

• Design and improve retrieval-augmented generation (RAG) pipelines combining semantic and keyword search, including graph-based RAG (GraphRAG) approaches for multi-hop and relationship-aware retrieval
• Build and fine-tune embedding and retrieval strategies for domain-specific text
• Work on OCR and document-understanding pipelines for scanned documents and PDFs
• Improve search relevance and reduce hallucination in AI-generated outputs
• Collaborate with backend engineers to productionize models as APIs
• Evaluate and integrate large language model providers, including hosted and self-hosted options
• Own accuracy and grounding of AI outputs, flagging low-confidence results appropriately

Must-Have Skills

• Strong Python; hands-on experience with LLM orchestration frameworks (e.g. LangChain or similar)
• Practical experience with vector databases and semantic search
• Experience designing and shipping RAG systems in production; familiarity with GraphRAG / knowledge-graph-augmented retrieval is a strong plus
• Familiarity with modern NLP/embedding model libraries
• Understanding of OCR/document parsing pipelines is a plus

Good to Have

• Experience in a high-precision or compliance-sensitive domain (legal, fintech, healthcare, etc.)
• Exposure to LLM evaluation, hallucination mitigation, and prompt engineering at scale
• Familiarity with common backend infra (relational databases, caching, containerized deployments)

Why Join: Work on a product where AI accuracy has real professional stakes for our users — not just another chatbot wrapper.