Software Developer, Data science, Machine learning

RITESH RANJAN

*/*/* ***** ******, New Delhi – ***031, Ph. No. – 821*******

**********@*****.*** · https://github.com/riteshranjan110

http://www.linkedin.com/in/riteshranjan11055

http://www.medium.com/@riteshk981

SUMMARY

I am a fresher having good understanding of machine learning concepts and data structures. I have worked on real world machine learning and deep learning projects which cover domains such as recommender system, taxi prediction, nlp, image segmentation, health care, time series data and malware detection. I have also taken part in two kaggle competitions . I am passionate about working on new projects. I am looking to start the career as an entry-level software developer and machine learning engineer with a reputed firm driven by technology.

EXPERIENCE

18/06/18 – 16/06/19

TRAINING, APPLIEDAICOURSE.COM

In this training we were taught machine learning algorithms and concepts. This course included classical machine learning and deep learning. We were taught KNN, Naive Bayes, Logistic Regression, linear regression, SVM, Random Forest, Boosting and Bagging. In deep learning we were taught MLP, CNN and LSTM.

16/03/19 – 16/07/19

INTERNSHIP, VAULTBOARD

In this internship I worked on resume parsing and resume matching. My responsibility was to extract data such as name, qualification, designation, previous company and experience of the candidate. The resume can be of file format such as .doc, .docx and .pdf. I also worked on the finding similar resumes given a query resume.

EDUCATION

Degree

College/School

Year of Passing

Aggregate

B.Tech (CSE)

AIACTR (GGSIPU), New Delhi

2019

86.67

12th

SM. Arya Public School, New Delhi

2014

92.40

10th

St. Paul’s School, Begusarai

2012

91.80

SKILLS

Programming Languages

C, Python

Database

MySql

Machine Learning

KNN, Naive Bayes, Logistic Regression, Linear Regression, SVM, Bagging and Boosting.

Deep Learning

MLP, CNN, LSTM, Bi-LSTM

PROJECTS

DEEP LEARNING PROJECTS:

JIGSAW UNINTENDED BIAS TOXIC COMMENTS CLASSIFICATION

The objective of this project was to classify whether a comment is toxic or not and

minimize unintended model bias.

This project is the motivated by the kaggle competition hosted by the Conversation AI

Team research initiative founded by jigsaw and Google.

Role:

1. Perform EDA and clean the data set given in the competition and vectorizing them according to the requirements of various models.

2.Training models such as logistic regression, Naive Bayes and testing them.

3.To develop various Bi-LSTM and Bi-GRU architectures and test them.

4.The best auc that we got was 0.9176 .

Technology Used: Python, textract, docx2text, sublime3, Spacy, pymysql

SEVERSTAL STEEL DEFECT DETECTION:

The objective of this project was to find whether the steel has any defect out of the

four defects given an image of steel. We had to maximize the dice coefficient.

This project is motivated from the kaggle competition hosted by Severstal.

Role:

1. To structure the data and prepare the images provided in the competition.

2.Since the images were of very high resolution we decreased the size of the images from 258x1600 to 128x800.

3.We used data generator so that we can train our model by passing images in batches instead of generating the whole training data set at once and then training. We did this because using the later method was giving memory error.

4.We have used various image segmentation models such as Unet with vgg16 and resnet34 as backbone.

5.We have achieved a dice coefficient of 0.8127 .

Technology Used: iPython notebook, anaconda, xgboost, sklearn, numpy, pandas

MACHINE LEARNING PROJECTS:

MICROSOFT MALWARE DETECTION

The objective of this project was to classify a given malware into 9 given categories

and to minimize log-loss

Role:

6. To perform EDA and use TSNE to check whether the datapoints can be classified into 9 classes or not.

7.To generate features from the given .byte and .asm files. I extracted unigram and bigram (top 1000 based on idf value) features from .byte files. Unigram features of asm files were used.

8.I also extracted 1000 bytes from image files of asm files which helped in reducing log-loss to 0.006.

Technology Used: iPython notebook, anaconda, xgboost, sklearn, numpy, pandas

TAXI DEMAND PREDICTION

The objective of this project was to predict the number of pickups that are going to

Happen in next 10 minutes in the region whose longitude and latitude are known.

Role:

1. To perform EDA and remove outlier points from the dataset.

2.To divide the whole dataset into clusters using clustering.

3.To generate features from the dataset and train and test models.

Technology Used: iPython notebook, anaconda, xgboost, sklearn, numpy, pandas

AMAZON APPAREL RECOMMENDATION SYSTEM

The objective of this project was to recommend most similar apparel given an apparel.

Role:

1. To perform EDA and clean the dataset by removing duplicate rows and rows whose columns are missing.

2.To extract features from the dataset which will be used in recommendation.

3.To build the system and test the results given by the system.

Technology Used: iPython notebook, anaconda, sklearn, numpy, pandas

SOFTWARE DEVELOPMENT PROJECTS:

RESUME PARSER

The objective of this project was to extract data such as name, qualification,

designation, previous company, email and phone number.

Role:

5. To extract raw text from different resume formats such as .doc, .docx or .pdf format.

6.To extract data such as name, qualification, designation, experience, company, email and phone number from raw text using regular expressions and Spacy.

7.To save the extracted data into excel file and in database.

Technology Used: Python, textract, docx2text, sublime3, Spacy, pymysql

RESUME MATCHER

The objective of this project was to find similar resume to a given resume.

Role:

1. To structure the unstructured data such as qualification details, job designation by categorizing them into similar categories

2.Creating rules so that resume are matched based on the experience of the candidate.

3.To extract skills from the resume by matching against a database of skills and using this to match resume.

Technology Used: Python, textract, docx2text, sublime3, Spacy

HOBBIES & ACTIVITIES

Loves to play cricket and watch movies.

Took part in organizing my college fest as coordinator in discipline and decoration team,

.

Contact this candidate