Overview
Work History
Education
Skills
Projects
Timeline
59
SANJEEV KHANNAN

SANJEEV KHANNAN

Jersey City,NJ

Overview

4
4
years of professional experience

Work History

Machine Learning Intern

KeeperAI
10.2023 - 12.2023
  • Developed Text-based Sentiment classification model using PyTorch and 3.5 million Amazon reviews, achieving impressive accuracy of 96% from scratch
  • Coordinated development of user personality trait classification system, prioritizing integration of pre-trained CLIP Vision Transformers from HuggingFace for detailed personality feature analysis
  • Implemented Continuous Integration/Continuous Deployment (CI/CD) deployment pipeline in AWS for model inference using Docker and Kubernetes, emphasizing scalability and stability.
  • Implemented natural language processing techniques for sentiment analysis and text classification projects.

Machine Learning Engineer

Zoho Corporation
08.2019 - 07.2022
  • Developed 5 impactful machine learning features for ZOHO CRM
  • Utilized statistical and AI/ML algorithms to enhance platform's functionality and user experience
  • Implemented feature-ranking algorithm using Apache Spark for data analytics and PostgreSQL for efficient querying
  • Designed ETL pipelines for seamless data extraction, transformation, and resulting in feature store for internal team use
  • Created custom data transformation pipeline optimized for efficient model training, data analysis, and seamless data export to dynamic dashboards, resulting in improved data visualization capabilities
  • Trained and deployed three chart recommendation ML models based on user activity data, which earned positive feedback from sales team
  • Led AutoML revamp that improved accuracy by 10% through pipeline optimization, feature selection, dimensionality reduction, predictive modeling, and advanced boosting techniques
  • Developed end-to-end model training pipeline by integrating Java with Spring MVC for efficient data retrieval and using Python with Flask and RabbitMQ for robust model training
  • Architected scalable and reliable systems using microservices and MVC design patterns, leveraging Apache Spark and Hadoop for high availability and fault tolerance in data processing.

Education

Master of Science - Computer Science

Pace University
New York, NY
05.2024

Bachelor of Engineering - Computer Science

Anna University
Chennai, India
04.2019

Skills

  • Programming Languages : Python, Java, C, Javascript Database: SQL(MySQL, PostgreSQL), NoSQL(MongoDB)
  • Machine Learning : Tree based Algorithms(Random Forest, XGBoost, LightGBM), SVM, Ensemble methods(Bagging and Boosting), NaiveBayes, KNN, Clustering(KMeans, DB-SCAN, Hierarchical Clustering); Recommender system
  • Deep Learning : Computer Vision (Object Detection, Image Segmentation, Pose Estimation, Face Recognition) and Natural Language Processing - LLMs [Retrieval Augmented Generation (RAG), Vector Databases - (AnalyticDB, ChromaDB), Quantization,]
  • Libraries and Frameworks: Tensorflow, Pytorch, SKLearn, Transformers, ONNX GenAI: LangChain with OpenAI APIs, VertexAI, RabbitMQ, Kafka
  • Cloud Platforms: AWS(EC2, S3, RDS, SageMaker), GCP
  • Version Control: Git, GitHub
  • Big Data Technologies: Apache Spark, Hadoop, MapReduce
  • Build and Deployment Tools: Docker, Kubernetes, Maven, Jenkins

Projects

  • Generative Multi-Modal Image Captioning - Developed an image captioning model using a multi-modal cross-attention mechanism. Leveraging approximately 70 million parameters, InceptionV3 for feature extraction, and Bidirectional LSTMs for sequence modeling, the model is trained on a dataset of 30,000 images from the Flickr Image Dataset, each with 2 captions. Achieved a BLEU score of 2.9, generating contextually relevant captions for images.
  • Product Reviews Sentiment Prediction using LSTM - Developed an Amazon Reviews text Sentiment Prediction model utilizing LSTM neural networks, trained from scratch on the Amazon Reviews Dataset of 3.5 million reviews. Created a custom data pipeline to load data in batches during training and testing, experimented with different layer architectures, resulting in an accuracy of 96%.
  • HandsJointDetection using X-Ray images - Developed a Hand Joint Detection project to accurately identify key joint positions in right-hand X-ray images. The dataset was meticulously preprocessed to exclude images without labels or with both hands/left hand. Utilizing convolutional layers for feature extraction and fully connected layers for regression, the model predicts 36 values per image, including x and y coordinates and finger angles for 12 key points. This project aims to improve medical diagnostics and treatment planning for hand conditions.

Timeline

Machine Learning Intern

KeeperAI
10.2023 - 12.2023

Machine Learning Engineer

Zoho Corporation
08.2019 - 07.2022

Master of Science - Computer Science

Pace University

Bachelor of Engineering - Computer Science

Anna University

Similar Profiles

CREATE PROFILE
SANJEEV KHANNAN