Education

****-**** ****** ** ******* GPA *.** (V. Good) University of Hamburg, Germany

Intelligent Adaptive Systems Major

****-**** ******** ** ********** 8.0/10 Jaypee University, India Computer Science Major

2007-2008 High School 84.8/100 Rewa, India

Specialized in mathematics, physics & chemistry

2005-2006 Higher Secondary School 85.6/100 Rewa, India Specialized in mathematics & science

Experience Summary

2 years of deep learning and 7 years of rich software engineering experience across di erent technology.

• Worked on di erent computer vision tasks such as pedestrian detection and tracking, semantic segmentation, saliency prediction through transfer learning, visualization of deep networks, tweaked pre-trained CNN models(VGG16, VGG19, Xception, ResNet50) and used OpenCV to solve specific problems.

• Experienced in implementing various model-free reinforcement learning algo- rithms such as Q-learning, SARSA, policy gradients in a manner that can solve the problems formulated as MDP.

• Worked on continuous delivery and continuous integration tools such as jenkins, Mesosphere/ Marathon, cron jobs using DC/OS and docker for platform agnostic applications.

• Proficient with HPC frameworks(for instance SLURM, LSF) used for e cient shared GPU cluster(s) management across team.

• Writer at Towards Data science medium community and write posts on AI/ML/DL topics. I publish my blogs intermittently at nilesh0109.medium.com Work Experience

Mar 21- now Machine Learning Engineer Langtec, Hamburg(DE) Leveraging self-supervised learning for improving semantic segmen- tation model trained on cityscapes dataset. Major focus of my work is on computer vision problems.

Apr 20-Feb 21 Research Intern Bosch Center for Artificial Intelligence, Renningen(DE) Working with "Robust and Explainable AI" group on representation learning. My Main focus is towards contrastive learning based meth- ods(BYOL, Simclr, MoCo etc) to learn robust representations which could be used for internal IOT data.

Jun 17-May 19 Software Engineer Dreamlines GmbH, Hamburg(DE) Supporting the data scientist with feature engineering and modelling the lead quality as a score submitted by customer (using Gradient boosted decision trees).

Enhancement and maintenance of existing Multilingual website for Eu- rope’s largest cruise booking portal along with A/B testing implemen- tation.

Oct 15-May 17 Frontend Developer Tacme LLC, Dubai(UAE) Developed a responsive Multilingual Website for one of the government clients of Dubai and supporting the website in webcenter based CMS integration.

Sep 14-Sep 15 Interactive Developer Sapient pvt limited Gurgaon(India) Worked on an e-commerce website of NIVEA.DE in the defect triage phase and focused on killing existing defect related to front end de- velopment so that the project can meet its tight timelines. Jan 12-Aug 14 Systems Engineer Infosys Limited, India Developed rich HTML5 based applications for US based Pharmaceuti- cal giant and successfully delivered 50+ applications. Technologies/ Tools Used: HTML5, CSS3, jQuery, AJAX, JavaScript, Veeva CLM library, SVG, XML.

Nilesh Vijayrania

Machine Learning Engineer

m nilesh0109.medium.com

m linkedin.com/in/nilesh-

vijayrania/

m github.com/nilesh0109

@ adk153@r.postjobfree.com

B Hamburg, Germany

T +49-152********

About me

I am a Machine development engineer

with strong focus on AI-powered

applications. My work is driven by my

keen interest in various artificial

neural network applications and allied

machine learning algorithms.

Technology wise, i majorly use python

and various of its scientific computing

packages(numpy, pandas, scipy,

matplotlib, seaborn) for EDA and

data modelling(scikit-learn, keras,

tensorflow, pytorch).

Skills

Languages: python,C, C++, JAVA, SQL,

JavaScript, PHP, HTML

Frameworks: tensorflow, pytorch, keras,

OpenCV, ROS

ML Libraries: pandas, numpy, scipy, mat-

plotlib, seaborn, sklearn

Backend Technologies: Nodejs, Php,

JSP

FrontendTechnologies:HTML5, CSS3,

JavaScript

WebDevelopmentFrameworks:Python

Django, ReactJS, AngularJS

Databases: mySql, mongodb

Others: Jenkins, Docker, AWS, Apache

mesos, queue

Past Projects

2020-now "Enhancing Label E ciency for semantic segmentation tasks using self-supervised learning" BCAI, Renningen

Using BYOL for pretraining on 20k unlabelled street view images in self-supervised manner, got 10% improvement in mIOU(0.65 to 0.74) on cityscapes val dataset. Used baselines architecture of FCN8s and deeplabv3plus with Resnet50 as backbone.

Link: https://github.com/nilesh0109/self-supervised-sem-seg 2020 Pedestrian Detection and Tracking

Using a side-view camera images over a flat floor, an object tracker is implemented in python using opencv. For pedestrian detection, Yolo- v3 and for tracking, kalman filter(SORT) is used.

Link: https://github.com/nilesh0109/PedestrianTracking 2019-20 Masters Project on "E ect of Non-verbal cues in Human Robot Inter- action" UHH, Hamburg

Through the carefully designed HRI experiment, which involved the interaction of a human participant with two NICO robots(programmed using ROS), we(masters project group) found participant’s slight incli- nation towards trusting the robot with non-verbal cues in comparison to robot with minimal non-verbal cues.

Link: https://github.com/nilesh0109/PHRI_1920

2019 Saliency Prediction System UHH, Germany

Used pretrained SalGAN on SALICON dataset to transfer learn and pre- dict the salient region of the input images. The dataset used was 1800 scene images collected through eye-tracking devices Link: https://github.com/nilesh0109/CV2_SoSe_19

2019 Chagas Parasite Diagnosis UHH, Germany

Diagnosis of chagas paracite disease using supervised learning data with GDA and SVM classifier

2018-2019 Visualization of deep networks UHH, Germany comparison of various visualization techniques for understanding complex behavior of deep networks(CNNs). The algorithms examined were deconvolution, guided-backpropagation, CAM, grad-CAM. 2018-2019 General Purpose Generative Chatbot UHH, Germany Experimented with Seq2seq architecture to generate adequate and in- telligent response for the general purpose chatbot. Implemented with Tensorlayer using LSTMs in encoder decoder architecture. Link: https://github.com/nilesh0109/seq2seqChatbot Achievements & Awards

• STAR OF THE BATCH Awarded with the title "STAR OF THE BATCH" during my summer internship under HPES at IIIT Allahabad, India from 10th June, 2011 to 15th july, 2011.

Certifications

2020 Deep Leanring Specialization Coursera

Applied Deep Learning Specialization

2019 Mathematics for Machine Learning Cousera

Linear Algebra Course by Imperial college london

2018 Python for Data Science for DataCamp Datacamp Data science Toolkit

2013 Microsoft Specialist Microsoft

Programming in HTML5 with JavaScript and CSS3

2011 Infosys Certified Android Developer Infosys

Android development

Interests: Deep Learning, Machine Learn-

ing Algorithms, Neural networks, Re-

inforcement learning, computer vision

tasks

Methodologies: Agile, Scrum

Source Code Management Tool: Git,

SVN

Languages

German

Hindi

English [The language scale is from 0 (Fundamental Aware- ness) to 6 (Expert).]

Hobbies Working out, playing sports, Tech podcasts and events, gadgets, AR/VR, Writing Tech blogs. My personal Space

(https://nilesh0109.medium.com/)

Contact this candidate