AI Research - Intern

BizNeuro AI; India; On-site; INR 223,200 - 390,600; Full time

Benefits offered by this job

Direct mentorship from industry experts

Opportunity to co-author research publications

Access to enterprise-grade GPU compute

Letter of Recommendation

Job summary

BizNeuro AI is seeking an AI Research Intern in India, focused on deploying large language models in enterprise environments. Candidates should have a background in AI/ML, with hands-on experience in model optimization and natural language processing. This role offers direct mentorship from renowned experts and the chance to publish research. Interns will tackle challenges related to hallucination control, LLM quantization, and document intelligence. The position provides significant impact on real production systems.

Qualifications

• Final-year or pre-final-year candidates preferred from IITs, NITs, or equivalent institutions.
• Hands-on experience with LLMs, especially fine-tuning or quantization.
• Strong foundations in machine learning and NLP.

Responsibilities

• Research and implement inference optimization strategies.
• Fine-tune open-source LLMs for BFSI-specific datasets.
• Implement hallucination detection and mitigation frameworks.
• Research and implement RAG architectures for fraud reasoning.
• Explore vision-language models for document tasks.

Skills

Machine learning

Deep learning

Natural language processing

Agentic AI frameworks

HuggingFace ecosystem

Python programming

Education

B.Tech / M.Tech / MS / PhD in Computer Science, AI/ML, Data Science

Job description

We're looking for AI researcher — someone who's spent time in the weeds of large language models and wants to apply that knowledge to a real production system solving real enterprise problems.

At BizNeuro AI, we build patented on-premise agentic AI for enterprise fraud prevention. Our agentic AI platform runs LLMs, SLMs, and ML models entirely within the client's infrastructure — which means the research challenges here are fundamentally different from standard cloud LLM work. You'll be working on inference efficiency, hallucination control, fine‑tuning, and multi‑agent orchestration in a production‑grade, privacy‑first environment.

If you want to publish, learn fast, and see your research ship in a live enterprise product — this is for you.

ABOUT BIZNEURO AI

BizNeuro AI is an enterprise agentic AI company building explainable, on‑premise fraud prevention technology for BFSI and e‑commerce enterprises. Our patented agentic AI platform runs ML models, Small Language Models (SLMs), and locally hosted Large Language Models entirely within the client's data boundary, ensuring 100% compliance, zero cloud exposure, and full audit transparency.

OUR SOLUTIONS

• DocNeuro — AI‑powered document & image forgery detection system
• MerchantNeuro — 360° merchant reputation & risk intelligence system
• EdgeNeuro — On‑prem/VPC agentic AI platform with agentic routing & hallucination control

Founded by Dr. Manish Gupta | PhD, Machine Learning — IIT Delhi | 4 US Patents | 14 Research Papers | 700+ Citations | One of India's Top 10 Data Scientists | NVIDIA's CEO Recognition, GTC 2020

ROLE OVERVIEW

We are looking for a highly motivated AI Research Intern with a strong background in large language models, model optimization, and applied NLP research. The intern will work directly with the founding team on research challenges central to BizNeuro AI's EdgeNeuro platform — including on‑prem LLM deployment, quantization, fine‑tuning for domain‑specific tasks, hallucination mitigation, and agentic AI orchestration.

This is a high‑ownership, research‑first role. Work produced during the internship may be published, patented, or deployed directly into production.

RESEARCH AREAS & RESPONSIBILITIES

1. LLM Quantization & Inference Optimization: Research and implement inference optimization strategies.
2. Fine‑Tuning & Domain Adaptation: Fine‑tune open‑source LLMs on BFSI‑specific and document intelligence datasets. Implement PEFT techniques for compute‑efficient adaptation.
3. Hallucination Control & Output Reliability: Research and implement hallucination detection, mitigation, and monitoring frameworks for production LLM pipelines.
4. Agentic AI & RAG Architectures: Research and implement RAG architectures for document‑grounded fraud reasoning and explainability.
5. Multimodal & Document AI (Desirable): Explore vision‑language models (VLMs) for document forensics tasks.

ELIGIBILITY & REQUIRED SKILLS

Academic Background

• Final‑year or pre‑final‑year B.Tech / M.Tech / MS / PhD in Computer Science, AI/ML, Data Science, or equivalent.
• Candidates from IITs, NITs, IIITs, or equivalent institutions preferred.

Technical Skills (Required)

• Strong foundations in machine learning, deep learning, and natural language processing.
• Hands‑on experience with agentic AI frameworks.
• Proficiency with HuggingFace ecosystem (Transformers, PEFT, Datasets, Evaluate, Accelerate).
• Demonstrated practical experience with LLMs — at least one of: fine‑tuning, quantization, RAG, evaluation, or prompting pipelines.
• Comfortable reading, understanding, and implementing from research papers.
• Strong Python programming and software engineering fundamentals.

Research Orientation

• Ability to independently scope, execute, and document research experiments.
• Intellectually curious with a bias for rigour and reproducibility.
• Proactive and self‑directed — able to work with minimal supervision.

WHAT WE OFFER

• Direct mentorship from Dr. Manish Gupta (PhD IIT Delhi · One of India's Top 10 Data Scientists · 4 US Patents · 700+ Research Citations).
• Opportunity to co‑author research publications, technical reports, or patent filings.
• Real production impact — work ships into a live enterprise AI platform.
• Letter of Recommendation based on performance.
• Access to enterprise‑grade GPU compute for research experiments.