Snehanshu Mukherjee

<Solving real problems with ML>
Dhanbad, India.

Education

Indian Institute of Technology (ISM) Dhanbad

→

Bachelor of Technology

Electrical Engineering & Data Science

Grade: 9.00/10

Courses

OOPs (C++), Building Software Systems, Data Structures & Algorithms, Data Mining, DBMS, Artificial Intelligence, Internet Technology, Prob & Stat, Linear Algebra

Work

NeurAl Lab, TU Eindhoven
|

Research Intern

Summary

Research on LLM guided sparsity and learning agents.

Highlights

Working on MoE & sparse attention schemes

Axis Bank Business Intelligence Unit
|

Data Science Intern

Summary

Worked with Personalisation Team to build in-house recommendation engine.

Highlights

Designed & built Axis Bank's first credit card Recommendation System for 4.3Cr customers.

Developed offer merchant affinity model capability for personalized offers.

Coded custom matrix operation functions for Spark DataFrames increasing performance of models 3X

CANDLE Research Lab, IIT Roorkee
|

Research Intern (Computer Vision)

Summary

Research work on Object Detection and related loss formuations.

Highlights

Developed lightweight object detection model (~ 3M params) for Alberta Construction Image Dataset (ACID).

Achieved 0.71 mAP i.e 7% over baseline mAP, beating YoLoV8 on construction datasets.

Projects

RakuMon

Summary

Developed multi-agent framework for hyper-personalized e-commerce shopping Implemented few-shot prompting for agents and image search mode. Built a generative agent for customized product descriptions based on user context.

StoryGPT

Summary

Decoder Based Small language model, coded from scratch. Trained over small story dataset (TinyStories). Finetuned on subset of Wildchat.

T_lib

Summary

C++ library to perform linear algebra on strided tensors CUDA C kernels for basic DL operations @ cuda_sensei Aims to support automatic differentiation and allow neuralnet implementations

Achievements & Open Soure

Contributed to Maya (multilingual multimodal AYA)

Awarded By

Cohere for AI

Part of AyA expedition'24. Helped integrating SigLip as vision encoder and wrote eval script to calculate BLEU score.

10th team out of 3k+teams at Fibe Hack The Vibe 2024

Awarded By

Fibe

Built Newsense - BERT based text classifier for news articles.

2nd runner up at Rakathon 2024

Awarded By

Rakuten India

Annual GenAl innovation challenge organised by Rakuten India.

PyTorch Docathon'23 Contributor

Awarded By

PyTorch

Fixed docstring errors in torch.nn.functional and torch.cuda\, torch.optim\

Publications

ConstructNet: A Deep Learning Object Detector for Construction Site Surveillance

Published by

International IEEE Applied Sensing Conference (APSCON)

Summary

in proceedings of 2nd International IEEE Applied Sensing Conference (APSCON), 2024

Skills

Languages & OS

C, C++, Python, CUDA, MySQL, Linux.

Libraries & Tools

PyTorch, CUDA, PySpark, NumPy, AWS, Scikit-Learn, NLTK, Einops, Milvus, HF, W&B.