AAEAAQAAAAAAAAOKAAAAJDhhN2QzOWQ2LTNhMTktNDgwNS04MjBhLTdhNTc4ZWM5MmQxZA.jpg

Naveen Benny

Deep Learning & AI Consultant

Summary


  • Deep Learning/AI consultant working on Computer Vision and NLP
  • Passionate about reading up on the latest research in the field and implementing papers from scratch especially ones on GANs. You can find a few of my implementations on Github (Link Above).
  • Highest Rank 88 on a Kaggle competition - Invasive Species Monitoring - Used modern state of the art architectures to develop algorithms to accurately identify whether images of forests and foliage contain invasive 'hydrangea' or not

Skills


Algorithms - 
Deep Learning | Computer Vision | NLP | Classification | Image Segmentation | Object Detection | CNNs | RNNs | LSTMs | Generative Adversarial Networks - Progressive growing of GANS | Self Attention GANs | Wasserstien GANs | Conditional GANs | CycleGANs | Autoencoders | Experience with SOTA architectures DenseNet, ResNet, Inception etc. | Machine Learning | Simulation | Decision Trees and Forests | SVM | Regression | Time Series Forecasting | Clustering algorithms | PCA

Languages & Libraries - 

Python: For research - PyTorch| For production - Tensorflow |  Theano | Numpy | Scikit-learn | OpenCV | Matplotlib |  SciPy 
Other: Torch | Caffe | Octave | R Programming | SQL | MongoDB

Domains - 

Computer Vision: GANs | Classification | Object Detection | Image Segmentation | Pose Estimation
NLP: Multimodal Image captioning | Question Answering | Word Embeddings |  Seq2Seq Models | Generative models | Text Summarization
Other: Fraud & Risk | Marketing and Campaign Optimization | Airline Operations

Courses


stanford.png

Convolutional Neural Networks for Visual Recognition

CS231n - Stanford University 

coursera-squarelogo-1483740136459.png

Machine Learning by Andrew Ng

Coursera - Stanford University

https://www.coursera.org/learn/machine-learning


fast.jpg

Cutting Edge Deep Learning For Coders

Fast AI

http://course.fast.ai/

Experience

Deep Learning Consultant |Self-employed | Jan'17 - Present

Description: Working with clients helping them implement deep learning based solutions.- 
Projects:  
  • Document Segregation - Built a deep learning based document segregation pipeline to classify documents based on their content using a ResNet18 
  • Subjective Grading - Used AWD LSTMs (Merity et al.) to grade student (grade 5) subjective short answers based on reference answers 
  • Document Data Extraction - Used Image processing and NLP algorithms to extract required data from unstructured documents 
  • Predictive Maintenance - Built an LSTM based solution to predict mechanical failures using sensor data
logo-1200x630.png

Data Scientist | Freecharge Payment Technologies | Oct'15 - Jan'17

Description: Worked as an individual contributor in the data science team helping build large-scale machine learning systems


Projects: 

  • Fraud & Risk: 
    • Built a scalable fraud management system using Machine Learning algorithms to detect fraudsters based on device, account, email, behavior data
    • Built a risk engine using random forest and neural nets to score and block users based on behavior and device patterns 
  • Marketing & Campaigns:
    • Helped predict user churn for proactive targeting using reactivation campaigns
    • Detect patterns in customer journey and helped devise effective strategies to acquire and retain good quality users
    • Segmented users based on their lifetime value that helped design customized campaigns 

freecharge-10-1486720812.jpg

Decision Scientist | Mu Sigma Business Solutions | Nov'13 - Oct'15

Description: Worked as team lead solving long-term analytics and optimization problems for largest domestic airlines in the U.S. Also responsible for managing client relationships, client workshops and mentoring of juniors.

Projects:
  • Forecasting Funds Expiration: Helped client pre-book a revenue of ~$400M on Wall Street by forecasting fund expiration a year in advance using time series models
  • Airline On-Time Performance: Built a system that used monte-carlo methods to simulate Airline Network to help improve and predict on-time performance 
  • Airport Segmentation: Helped segment airports using clustering algorithms based on their operational characteristics to enable effective operational decision/strategy making
  • Optimized Fleet Introduction: Helped strategize introduction of newly acquired fleet optimally using decision trees
mu-sigma-logo.jpg

B.E Mechanical Engineering (Hons) | BITS Pilani Goa Campus | Aug'09 - May'13

  • Internship: Titan Industries, Hosur | Jun'12 - Dec'12

main-qimg-06d8bd78bc5b3a2f31388085718ae6a2.png

CBSE Class XII | St. Antony's Public School, Kottayam | Jun'07 - May'09

download (1).jpg

Extra-curriculars


  • Played for Mu Sigma football team winning numerous tournaments; voted Best player in many tournaments 
  • Co-founded BITSFC, a club that plays in the Goa football association’s (GFA) Second Division; was Vice-captain and leading goal scorer and played a pivotal role in its qualification from the Third Division in only its second season in the league
  • Runner-up in 3000m Middle East CBSE Zonal meet U16 category representing UAE
  • Winner of several college and school football and athletic events; captained the team to win around 5 internal football tournaments in college
  • Member of the ‘Department of Controls’ for Annual National Cultural Festival; organized various events during the fest; managed Christmas programs, cultural celebrations and sports events in college