Jul 2017 – May 2021 · 3 yrs 10 mos · Gurugram, Haryana, India

• Elevator Pitch
• · Deployed and optimized the algorithms for production level using Golang with proper connection handling, timeouts and queues
• · Used Supervisor to ensure the models kept running and get started in case of any failure
• · Implemented a Deep learning model (YOLO) to predict hand gestures by predicting the box centers and thereby tracking its location
• Live server maintenance
• · Deployed and maintained ML development codes on the live server with proper functionality and load optimization
• · Implemented proper queues, error logging and fault reports, mail service for immediate notification and timeouts for smooth API flow thereby ensuring no user gets affected
• Training Program
• · Prepared an exclusive training program for Machine Learning (ML) for freshers
• · Mentored each fresher to ensure smooth learning curve for the freshers so that they can work on any project independently
• Predictive Analytics for Resume Shortlisting
• · Developed a mathematical model to determine weights of detected features in a Resume from historical data
• · Designed a scoring algorithm to predict likelihood of a Resume getting shortlisted in any particular company
• · Extracted major insights for a company by detecting thick and thin patterns in weights of detected features

Jul 2016 – Jul 2017 · 1 yr · Gurugram, Haryana, India

• Categories Detection in a Resume using NLP
• · Led a team of 4 and implemented machine-learning models for detection of categories inside a Resume thereby aiding in recognition of type of written content allowing the product to give feedbacks on low content resumes
• · Developed testing panels for conducting thorough error-analysis of machine-learning models to improve accuracy
• · Designed tagging panels for storing and generating input data to train machine-learning models
• · Deployed CNN and Logistic Regression using Tensorflow & Sklearn for detection
• Matching Algorithm for contextual similarity detection
• · Developed a similarity matching algorithm using pre-trained sense2vec vectors and spacy POS tagger to output contextually similar samples from database to given bullet thereby vastly improving bullet-level feedbacks
• · Designed an architecture to optimise required time by 1/10th to compute similarity scores for given bullet
• · Extended application of algorithm across different products according to their use-case
• Elevator Pitch
• · Deployed Video Processing, Face-detection and Object-detection algorithms and Classification Models using DLIB, OPENCV and Keras for evaluating physical behavior of an individual in an interview
• · Developed and implemented algorithms to detect Eye Movements, Smile and apt Body Posture using facial points
• · Employed CNN and deep ANN machine-learning classification models to detect Tie, Suit and Beard from image
• · Peformed audio analysis and recognized filler words, pauses; trained a speech-to-text model for evaluating speaking skills of an individual in an interview.

May 2015 – Jul 2015 · 2 mos