Ashish Jindal

Background

About Ashish Jindal

Passionate Data Scientist with strong Computer vision skills, 7 years experience in working with FinTech, Automotive, Insurance and Finance industry. Skilled in Python, Computer vision, Generative AI and ML/AI. Strong, Dynamic, and composed AI professional with a bachelor's degree in CSE, Data science specialization, and relevant course certificates. I can do : Object detection Image classification Scaling infrastructure - Kubernetes AI/ML Product Development Research paper implementation Finantial Fraud Detection Rest API creation (Django) Model Deployment NLP-based projects Industry Projects: Architecting CLIFF: A Scalable Image Search Engine: • Created CLIFF, a scalable image search engine utilizing deep learning embeddings and powered by Qdrant for efficient similarity searches via cosine distance. • Developed a highly scalable image search platform in Python leveraging Azure Kubernetes. • Engineered solutions following agile methodologies, prioritizing clean code, object-oriented programming (OOP), and design patterns. Technologies employed include FastAPI, Kubernetes, Azure cloud suite, Azure service bus, Terraform, Postgresql, and Qdrant. KYC Automation: • Detection: Developed object detection model to detect KYC card using SSD • OCR: Implemented different solutions like Tesseract, Google vision, Azure vision OCR • Extraction: Mapping using various regex/logic to extract key entities • Model Deployment: Used Tensorflow Serving to deploy various models Face liveliness and Face matching: • Filter out non-live (Photo of a photo, mask on face, video replay) from face images • Used ensemble learning with various image classification models Document Corner detection: • Real-time corner detection by recursive application of a CNN • Implemented a research paper to replicate CamScanner • Executed end-to-end pipeline from data collection to implementation of novel CNN architectures in Keras. Car damage detection using Deep Learning: • Trained an end-to-end image-based car damage detector with feedback learning capability for automation in insurance industries Bank Transaction Categorization: • Labelled bank statement transactions for credit underwriting decisions. Tools & Technologies: Python Deep Learning(DNN/CNN/RNN) Anaconda/PyCharm/Jupyter Notebooks Tensorflow/Keras/pandas/sk-learn/matplotlib CUDA/Cloud Computing Ensemble – Bagging/Boosting(XGboost, Adaboost, GBM) Hyperparameter Tuning/Model Selection Feel free to connect with me at: ashishjindal1997@gmail.com +91-8107259905