Experienced Senior Data Scientist with expertise in data analysis, machine learning, and statistical modeling. Proven leadership in mentoring teams, developing data strategies, and aligning data science initiatives with business objectives. Skilled in managing complex projects, delivering high-quality results on time and within budget, and driving data-driven decision-making across organizations.
Designed and implemented a CNN-based auto-encoder model to detect Account Takeover (ATO) fraud by analyzing user online activity, device information, and domain-specific features, significantly reducing the volume of investigation cases.
1) Invoice Parameter Extraction: Employed image processing and NLP (NER models) to automate invoice parameter extraction, improving data accuracy and efficiency.
2) Employee Performance Monitoring System: Implemented a system that classified user screen recording sessions to enhance employee performance monitoring and management decisions.
3) Email intent detection system integrated to salesforce: intent like complaint, feedback, query, compliment.
Machine Learning(LR,SVM,bagging and boosting)
Deep Learning (CNN, LSTM, ANN, GNN)
NLP(Language models)
Personalised recommendations(graph neural networks)
Model evaluation and dashboard implementation
Keras, Tensorflow, Pytorch
LLM (LLama,Falcon open source models))
Image processing
Statistical analysis and inference
Python3, Django, Docker, pyspark
Team management
Manage multiple projects
Define quarterly targets and key results
Machine Learning: statistical models , feature selection using correlation and data normalization, transforming data, hyper parameter tuning, deciding loss functions and optimizers, ensemble models , bagging and Gradient boosting methods,PCA(principle component analysis)
Deep Learning: CNN , activation function, different layers, , regularization techniques, data preparation, data preprocessing, training models and
NLP: NER(named entity recognition), sentiment analysis, word2vec,CBOW,SKIPGRAM,Glove, BERT, TFIDF, N-gram,Transformers, Fast text embedding(char embedding).
Validation Techniques : AUC (Area under curve analysis) / ROC ( receiver operating characteristic curve),confusion matrix, Precision, Recall, F1 score.
Project : personalized recommendations to the user based on user history and demographics, using Graph neural networks (deep graph library).
Project : detecting products,shelves and positions in Retail store, using deep neural network
Project: Extracting invoice parameters , invoice number, date,account number,vendor,ship,and remit addressed
Project: extract contact entities , person name, title and company name email address from email threads and populate these details into sales force contact management screen
Project: Build, person ,facemask detection and social distance monitoring Video analytics application and deployed in SHINOBI CCTV monitoring app, Live Detection and streaming
Project : Text classification , intention detection in email threads, is email contains complaint, feedback, query, compliment, information.
Project : user activity monitoring tool , deep neural network to monitor user system , how much time spent on which application, browser , command line, media player, excell
Project : text classification, monitor user screen and classify the amount of improper text present on screen , to measure how much a child is exposed to abuse,hatred on social media platforms
Neural Networks And Deep Learning: coursera.org/verify/SLTJ3EGKFWQQ
Neural Networks And Deep Learning: coursera.org/verify/SLTJ3EGKFWQQ