Skills

• Amazon Web Services (AWS)
• TensorFlow | Keras | Media Pipe
• Python | REST APIs
• Jupyter Notebook | Google Colab
• OpenCV | Raspberry Pi
• Numpy | Pandas | Scikit-Learn | Matplotlib
• Tableau | KIbana | Elastic Search
• Alexa | ChatGPT | Meta AI | Gemini

ML Models

• Convolutional Neural Nets (CNNs)
• Transfer Learning
• LSTM | RNN
• Auto-Encoders | Style Transfer
• Large Language Models (LLMs)
• Generative AI | GANs

Summary

• Hands on experience building practical AI applications
• Applying Deep Learning, Machine Learning algorithms on Multimedia datasets containing Audio, Video, Image, Text, Dialogs
• Leadership Roles: Sr Manager at Cognizant | Associate Director at Happiest Minds | Principal Data Scientist at Make My Trip | Project Lead at Persistent
• Notable Clients: ESPN, Warner Music, Godrej
• Key Areas: Computer Vision, Deep Neural Networks, Image Recognition, Video Understanding

Academics

• University of Minnesota, Masters in Computer Science
• Pune University, Bachelors in Computer Engineering (Best Outgoing Student Award)
• Research Intern at Sony – Tokyo, Japan | Summer Intern at Amazon.com – Seattle, WA | Summer Intern at University of Southern California (USC) | Research Staff at Singapore Management University – Singapore
• Research Paper on Humor Analysis in FRIENDS Dialogs at EMNLP, Sydney

Job Functions

• Mentoring & guiding junior team members, to resolve any issues or challenges faced during development and deployment
• Preparing & presenting project proposals by outlining the scope of work, project timeline, budget, team structure, solution architecture and technology stack
• Design Solution Architecture by researching latest coding platforms, tools & libraries to solve given use-cases efficiently & accurately
• Work closely with business development, sales, marketing teams to ensure projects are aligned with overall corporate strategy
• Build products & services that showcase creative innovative ideas & latest technology trends

Projects

• Media Pipe: Extract key landmark points using Media Pipe’s hand, face mesh and body pose APIs. ML model is then built using scikit-learn and neural network in keras to
  • (a) capture facial expressions (happy, angry, scared, sad etc) using facial landmarks
  • (b) detect yoga poses (Warrior, Mountain, Triangle, Wheel Pose etc) using full-body skeleton posture (knee, shoulder, ankle joints, spine etc)
  • (c) detect hand gestures like 👍 👎 👌 ✌️ 🤘
  • (d) detect activity in sports videos (cycling, skating, jogging, climbing, boxing etc) by analyzing body posture of players across multiple video frames using sequential model based on RNN
• Transfer Learning: generic transfer learning based ML model that can be easily applied to variety of datasets, to solve number of image recognition tasks and use-cases, without changing a single line of code:
  • (a) Food Classification: classify food images by identifying type of dish (pizza, burger, noodles, sandwich etc)
  • (b) Landmark Detection: identify famous historic landmarks and architectures (Taj Mahal, Burj Al Arab, Hampi Stone Chariot, Jaipur’s Hawa Mahal etc) in photos
  • (c) Optical Character Recognition (OCR): recognize indian language characters in hand written text
  • (d) Face Recognition: identify famous Indian celebrities (politicians, bollywood actors, sports figures)
  • (e) Cartoon Characters: detect popular cartoon characters like Mickey Mouse, Simpsons, Minions, Hello Kitty, Spider-man, Batman etc printed on kids articles like backpack, water bottle, pillow, t-shirts
  • (f) Wild Life: identify animals (gorilla, elephant, lion, tiger, giraffe, deer) in wild life documentary films
• Auto-Encoder: encoder – decoder model to to colorize B&W images
• Computational Music: Audio classification model to
  • (a) detect musical instruments played in the audio (violin, guitar, flute, piano etc)
  • (b) detect musical notes (C, C#, G, A#, F etc) played on piano
  • (c) voice recognition to extract lyrics (words) from songs