Engineer · Researcher · Builder

Arnab Datta.

Electrical Engineering undergraduate at IIT Kanpur, building intelligent systems at the intersection of machine learning, robotics, vision and science.

Scroll to discover
01 · About

Curious by design.

I like problems that refuse to stay in one discipline—where models meet machines, scientific data becomes useful, and ambitious ideas have to survive contact with the real world.

What drives me

I build end-to-end systems: from training the model and designing the algorithm to making it fast, reliable and genuinely useful.

Currently

Leading Team ERA

Heading a 19-member undergraduate robotics team and working on autonomous systems, 3D vision and predictive control.

02 · Competitions

Research with results.

Two international-scale challenges. Two very different scientific domains. The same obsession with making complex models perform under real constraints.

COMPETITION · 01 🥈 Inter IIT Tech 14.0
Remote sensing · Multimodal AI · 2025

ISRO Geospatial

An intelligent multimodal chatbot for satellite imagery, enabling captioning, grounding and visual question answering across RGB, FCC and SAR images.

  • Orchestrated Qwen3-VL, YOLO, SAM and RemoteCLIP through LLaMA 3.1.
  • Outperformed SOTA on grounding and float queries with sub-10-second inference.
  • Accepted for oral presentation at IEEE IGARSS 2026, Washington DC.
View project on GitHub ↗
COMPETITION · 02 🥉 CAFA 6 · Kaggle
Bioinformatics · Foundation models · 2025–26

Protein Function Prediction

A high-performing pipeline for predicting Gene Ontology terms from amino-acid sequences, designed to improve sparse-label generalization.

  • Generated protein embeddings using the 650M-parameter ESM2 model.
  • Combined KNN retrieval with GCN-enforced Gene Ontology hierarchy constraints.
  • Improved weighted F1 from 0.194 to 0.366; ranked 210 of 2,259 teams.
View competition credential ↗
03 · Papers

Accepted research.

Two accepted papers spanning remote sensing, robotics and perception. ORCID iD: 0009-0000-6060-6248.

IEEE IGARSS 2026 · Washington DC · Co-author

TRINETRA: Text-guided Remote-sensing INterpretation via Ensemble-based TRAnsformers

Accepted for oral presentation at the IEEE International Geoscience and Remote Sensing Symposium, connecting multimodal AI with satellite imagery understanding.

RoboCup Symposium 2026 · Springer LNCS/LNAI

Visual Localization and Detection via Monocular Omni-Vision

Accepted at the 29th RoboCup International Symposium in Incheon, Republic of Korea, focused on cost-effective monocular perception for autonomous mobile robots.

05 · Leadership

Making teams click.

Technical depth matters. So does creating the environment where people can do their best work together.

Apr 2026 — Present

Team Head, Team ERA

Science & Technology Council · IIT Kanpur
  • Leading a 19-member undergraduate robotics team across technical and operational work.
  • Part of the first and only Indian team to qualify for RoboCup MSL 2026.
  • Collaborated with 20+ science and technology teams for the SnT Pavilion.
May 2025 — May 2026

Secretary, Programming Club

Science & Technology Council · IIT Kanpur
  • Maintained technical roadmaps, blogs and projects for the campus community.
  • Led Spring Camps in Machine Learning and Web3 for first-year students.
  • Built interest groups around emerging domains and peer-led learning.
07 · Toolkit

Foundations & fluency.

A strong mathematical and engineering base, paired with the tools to move comfortably from an experiment to a deployed system.

Education
2024 — PRESENT

B.Tech., Electrical Engineering

Indian Institute of Technology Kanpur · CPI 9.5/10

2024

ISC · Class XII

Don Bosco School, Kolkata · 98.25%

2022

ICSE · Class X

Don Bosco School, Kolkata · 99.2%

LanguagesPython · C · C++ · Java · Bash · JavaScript · Solidity · Assembly
ML & DataPyTorch · TensorFlow · NumPy · Pandas · Matplotlib · OpenCV
SystemsROS2 Humble · NVIDIA Jetson · Arduino · Git · Fusion 360
PlatformsKaggle · GitHub · Hugging Face · React.js · LaTeX
CourseworkMachine Learning · Statistical NLP · Signals & Systems · Probability & Statistics · Analog Electronics · Electromagnetism

Let’s build something ambitious.

Open to research collaborations in ML, robotics and vision, competitive research problems like CAFA or IGARSS-style challenges, and robotics teams looking for a perception/controls lead — or an idea that’s just weird enough to be worth trying.