Shuvam Banerji Seal

BS-MS · Chemistry Major · CS Minor

Indian Institute of Science Education and Research, Kolkata

Research experience

July 2025 to present

A Hybrid RAG Architecture for Verifiable Answer Synthesis with Automated Source Citation

RAG systems · Dr. Dwaipayan Roy, CDS, IISER Kolkata

Comprehensive RAG pipeline for IISER-K intranet: automated data acquisition via Selenium, OCR with PyMuPDF/EasyOCR, context-aware chunking, Qwen3-4B embeddings in FAISS, hybrid BM25+dense retrieval, hypothetical-answer query refinement, and verifiable inline citations in Streamlit.

  • RAG
  • FAISS
  • Qwen3-4B
  • BM25
  • Selenium
  • Streamlit

2024

Developing an Advanced Retrieval Model for TREC-Tip of Tongue Queries

Published · Information retrieval · Dr. Dwaipayan Roy, IISER Kolkata · TREC 2024 Proceedings

Multi-layer BM-25 filtering with dynamic search-domain contraction, transformer query expansion, semantic matching via Local LLMs with multi-shot chain-of-thought prompting, matching DPR-class performance.

  • BM25
  • Query Expansion
  • Chain-of-Thought
  • DPR

2024–25

Computational Modeling of [VO(SALIEP)(DTP)] as Water Reducing Catalyst

Computational chemistry · DFT · Dr. Soumyajit Roy, IISER Kolkata · To be submitted

DFT methods (B3LYP) for MO energy calculations and electronic structure analysis in Gaussian; mechanistic studies via transition state theory and reaction pathway analysis for catalyst performance prediction.

  • DFT
  • Gaussian
  • B3LYP
  • Transition State

Publications

FIRE 2025

Hierarchical Opinion Classification with Large Language Models

Submitted · NLP and LLMs · With Alok Mishra and Utkarsha Ghosh

Parameter-efficient fine-tuning of Gemma-1B for hierarchical opinion classification: lightweight two-layer head, selective transformer block updates, 8-class reformulation with class-weighted loss for minority-class performance.

  • Gemma-1B
  • PEFT
  • Opinion Mining

2024

IISERK@ToT_2024: Query Reformulation and Layered Retrieval for Tip-of-Tongue Items

Information retrieval · TREC 2024 · With Subinay Adhikary, Soumyadeep Sar and Dr. Dwaipayan Roy

Four-step query reformulation + two-layer retrieval (BM25 + LLM), achieving best NDCG metrics for the TREC 2024 Tip-of-the-Tongue track.

  • BM25
  • NDCG
  • TREC 2024

Projects

Aug 2025

IndicAgri: AI platform for Indian agriculture

Hackathon · Capital One Launchpad 2025 · Top 14 / 5000+ Teams

Multi-modal (text+voice) RAG chatbot for 20+ Indian languages with Gemma/DeepSeek LLMs; agentic pipeline for 15k+ document curation.

  • RAG
  • Gemma
  • DeepSeek
  • Indic NLP

2025

Automated WhatsApp Bulk Messenger with Google Sheets Integration

Independent · Automation

Python + Selenium bulk messaging with CSV/Google Sheets API, Unicode, session persistence, and error handling.

  • Python
  • Selenium
  • Google Sheets API

2025

Automated Event Coupon Management System with QR Verification

Independent · Event tech

Flask + Google OAuth 2.0, bulk email, unique QR codes, real-time mobile verification, and automated thank-you emails.

  • Flask
  • QR Codes
  • Google OAuth

Feb 2025

RAG Application for CCA-2015 Legal Document Retrieval

Independent · Legal tech

Selenium scraping, then Mistral-OCR JSON base, ChromaDB, Nomic embeddings, and LLaMA-3.2:1B in Streamlit.

  • ChromaDB
  • LLaMA
  • Nomic

2024

GTK4 Full-Stack Application in C

Academic · Systems programming · Supervised by Dr. Kripabandhu Ghosh

Complete GUI with GTK4 / Glade, SQLite3 backend, and BM25-based NLP chatbot retrieval.

  • C
  • GTK4
  • SQLite3
  • BM25

2024

Wi-Fi Channel Optimizer

Independent · Networking

Bash automation for channel selection via network scanning + speed testing (speedtest-cli + nmcli).

  • Bash
  • nmcli
  • Network

Professional and leadership

2024–2025

Anicon 3.0 web lead, IISER Kolkata

Web development · Inquivesta XI

Developed and led web development for Anicon 3.0 at Inquivesta XI, IISER Kolkata.

2025

EFAML lab website, Dr. Soumyajit Roy

Research outreach

Designed the EFAML (Materials Science) lab page for Dr. Soumyajit Roy at IISER Kolkata.

2018–Present

Private educator and technical trainer

Education · Mentorship · 50+ students mentored

Advanced Computing and Basic Sciences instruction for board and competitive examinations.

2022–2025

Event organization and social impact

Community · Leadership

Organized Anicon 3.0 (500+ participants), GIAN Courses on Soft-Oxometalates and X-Ray Crystallography, Supra-Molecular Discussions 2024. COVID-19 relief coordinator; Ek-Pehal educational mentor. District Debate and Quiz finalist; Shotokan Karate; Art and Painting (stroke art and portraits).

Education

Expected 2027

BS-MS · Chemistry Major · CS Minor

IISER Kolkata

2021–2022

B.Sc Honours in Physics

Calcutta University

2019–2021

Higher Secondary (WBCHSE)

Jodhpur Park Boys' High School · 83%

2009–2019

Secondary (WBBSE)

The New Horizon High School · 83.75%

Technical expertise

Core proficiency

Python95%
Machine learning and AI90%
Information retrieval88%
Computational chemistry85%
C / C++82%

Programming languages

Python, C / C++, Java, Rust, Fortran, QBASIC

Scientific computing

NumPy, Pandas, SciPy, PyTorch, TensorFlow, Scikit-learn, Matplotlib, OpenCV

NLP and RAG

HuggingFace, LangChain, FAISS, ChromaDB, BM25, SpaCy, Selenium

LLMs and embeddings

Qwen3-4B, Gemma, DeepSeek, LLaMA-3.2, Nomic Embed, Mistral-OCR

Scientific software

LAMMPS, VMD, Gaussian, PyMOL, ChimeraX, AutoDock Vina

Development and tools

Django, Flask, GTK4, Qt, MongoDB, SQLite3, Git, Docker, LaTeX

Achievements and honors

Hackathons

2025

Capital One Launchpad: Top 14 / 5000+ Teams

2024

StatusCode1 GIAN: 1st Rank

2023

StatusCode0 MATLAB: 1st Rank

National competitions

2025

ChemEnigma: 1st Rank

2025

All Bengal Chemistry Quiz: 2nd Runners Up

2024

Mimansa: Zonal Topper

2023

NAEST: Zonal Runners Up

Scholarships

2023

Reliance Foundation Undergrad Scholar

2019

Best Young Scientist Speaker: Nanotechnology

Competitive exams

2022

JEE Mains and Advanced: Qualified

2022

IAT (IISER Aptitude Test): Qualified

2022

WBJEE: Qualified

Languages

English

Fluent

Bengali

Native

Hindi

Proficient

Full CV (PDF)