Greetings! I’m Amit Pandey — a researcher with a passion for building products, and a soft spot for poetry and teaching.

I am currently a Data & Applied Scientist at NetApp and a Ph.D. Scholar at IIIT Hyderabad.

My work sits at the intersection of multimodal AI, inference optimization, autonomous research, and agentic systems. I design, fine-tune, and productionize language and vision-language models — and the infrastructure around them — turning research ideas into systems that run reliably at scale. I’m equally interested in making models smaller and faster and more capable and autonomous (agents that can experiment, retrieve, and reason on their own).

Broadly, my interests span Generative AI, Natural Language Processing & Understanding, Computer Vision, Multimodal Learning, and trustworthy/secure AI.

I’m always happy to discuss new ideas or explore research collaborations — feel free to reach out.

Research Interests

• Multimodal AI & document/video understanding
• Inference optimization and efficient, low-resource model deployment
• Agentic AI — autonomous research, retrieval, and orchestration
• Representation learning, information retrieval, and trustworthy AI

Currently working on

• Natural language → production systems: a system that turns plain-language intent into complete, production-grade data pipelines (e.g., automatically composing Apache NiFi flows) to serve a range of business use cases (Personal Project).
• Agentic research infrastructure: autonomous experimentation, data curation, and evaluation with model-agnostic agents.
• Efficient multimodal models: pushing accuracy under tight latency and compute budgets.

Publications & Patents

• US Patent (Filed) — Artificial Intelligence Platform Security Mechanism: autonomous red-teaming and pre-ingest defense for AI data pipelines.
• ECCV 2026 — Open Your Eyes: Benchmarking the Detection of Fabricated Realities and Weaponized Ethics in VLMs.
• COLING 2022 — CitRet: A Hybrid Model for Cited Text Span Retrieval.
• NAACL SemEval 2022 — Transformer-based Architecture for Complex Named Entity Recognition.
• ACM SSDBM 2022 — Journey to the Center of the Words: Word Weighting Based on the Geometry of Word Embeddings.
• CODS-COMAD 2024 — Spatial Distributional Representation of Word Embeddings.

Education

• Ph.D., Computer Science — IIIT Hyderabad (In Progress) · Multimodal AI & Document Intelligence
• MS by Research, Artificial Intelligence — IIIT Hyderabad (2024) · Information Retrieval for Scholarly Document Processing
• B.Tech, Electronics & Communication Engineering — BVCOE, GGSIPU, New Delhi

Teaching

• Head Lab Instructor & Teaching Assistant — Foundations of Modern Machine Learning (FMML), IIIT Hyderabad.