Hitesh Kumar Gupta

FIRE 2026 Track Coordinator, IIT Kharagpur
Former R&D Intern, Dept. of Artificial Intelligence

Email· Google Scholar· Semantic Scholar· ORCID· GitHub· LinkedIn· Blog· CV

I recently completed an R&D Internship in the Department of Artificial Intelligence at IIT Kharagpur, working with Prof. Koustav Rudra. I continue to coordinate the cryptocurrency information retrieval track at the Forum for Information Retrieval Evaluation (FIRE 2026), organised by IIT Kharagpur.

Previously, I was an R&D Project Intern at IIT (ISM) Dhanbad, working with Prof. Sachin Tripathi, and before that a Deep Learning Intern at Birsa Institute of Technology, Sindri with Prof. Khustar Ansari.

In parallel with research, I worked four years as a software engineer at VaporFi Labs (2022–2026), a US-based fintech and Web3 company. I built event-driven indexers ingesting real-time data streams into PostgreSQL, REST and GraphQL APIs, and led end-to-end development of a marketplace handling 65k+ assets with a React/Next.js frontend.

I am largely self-taught. I started programming on a smartphone at 13, building chatbots for companies while in school, shipped a full-stack application from my phone at 17, and independently published my first research paper at 20, now cited across IEEE, IOP Science, and NCIT venues. I am a final-year B.Tech (Computer Science) student, open to research and engineering opportunities.

Research Interests

Responsible AI · Natural Language Processing · Information Retrieval · Computer Vision.

Publications

When Better Eyes Lead to Blindness: A Diagnostic Study of the Information Bottleneck in CNN-LSTM Image Captioning Models.
Hitesh Kumar Gupta. International Journal of Computer Applications, 2025. paper · doi:10.5120/ijca2025925560

Blog

Escaping the KV-Cache Bottleneck: vLLM vs HuggingFace Transformers on an H100: engineering notes on running ~100k LLM generations on a shared H100, and the gotchas in between. · All posts

Selected Engineering Work

nuqs-presets: production-ready React hooks for URL state (pagination, filtering, sorting), in TypeScript.
serverless-wiki-search: in-browser full-text search engine using SQLite WASM and HTTP range requests, with an automated data pipeline.
VaporFi/core-wagmi-connector: open-source wallet connector maintained under a company organisation.

Experience

R&D Intern, IIT Kharagpur May – Jun 2026

Dept. of Artificial Intelligence, with Prof. Koustav Rudra
R&D Project Intern, IIT (ISM) Dhanbad Feb – Apr 2026

Dept. of CSE, with Prof. Sachin Tripathi
Deep Learning Intern, BIT Sindri Jun – Jul 2025

Dept. of CSE, with Prof. Khustar Ansari
Software Engineer, VaporFi Labs 2022 – Apr 2026

Part-time · US-based fintech company (blockchain and DeFi domain)
- Built high-throughput event-driven indexers ingesting on-chain data streams into PostgreSQL in real time; composite SQL indexes held p95 query times under 200ms in sustained production load.
- Built REST and GraphQL APIs with Zod schema validation; TanStack Query caching and request deduplication cut redundant network calls ~40%.
- Led end-to-end development of a marketplace handling 65k+ assets (React/Next.js); raised Core Web Vitals from 63 to 95+ via an ISR/SSG refactor and selective hydration.
- Containerised services with Docker; set up CI/CD for self-hosted infrastructure.

Recommendations

Hitesh is one of the best developers, thinkers, and problem solvers, and would be an asset to any team he works with. We worked together at VaporFi, and he was always keen to solve problems outside of his domain and attentive to detail. He is a great person to work with.

Parth Patel — Senior Engineer, Yieldnest (prev. AAVE, Amazon)

His attention to detail stands out immediately. He catches issues most engineers miss, both in code and in design reviews. He also picks up unfamiliar systems faster than anyone I've worked with, which makes him reliable when thrown into new territory. Where he really shines is system design and execution. He thinks through edge cases up front, makes pragmatic tradeoffs, and follows through. The features he ships are stable, well-scoped, and rarely come back with regressions. Any team would be lucky to have him.

Gaurav U — Software Engineer, Mercor

Education

B.Tech in Computer Science & Engineering, K. K. College of Engineering & Management, Dhanbad. CGPA 8.5/10, first rank in programme. Expected 2027.

Awards & Service

Coordinator, cryptocurrency information retrieval track (organised by IIT Kharagpur), FIRE 2026.
arXiv Endorsement (cs.CV, cs.LG), obtained independently without institutional affiliation.
Hindustan Talent Honour, 2023.