Niveda Das. M

Generative AI, Machine Learning Engineer

Trivandrum

Summary

Results-driven Generative AI Engineer with hands-on experience in building, deploying, and mentoring deep learning and NLP projects. Proven track record in applying state-of-the-art models like Transformers, GANs, and LLMs to real-world use cases. Adept at research, development, mentoring, and integrating AI models into production systems. Passionate about driving innovation in AI through strong fundamentals and practical applications.

Overview

year of professional experience

Work History

Data Science Intern

Vlipper

12.2023 - 08.2024

Tech Used:

Built an AI model for lip-syncing Indian vocals using deep learning and computer vision techniques.
Conducted POC for Malayalam dubbing, showcasing multilingual NLP capability.
Developed a recommendation system using collaborative filtering and embedding-based methods.
Collaborated with the product team to deploy AI features into the platform using Python, TensorFlow, and Flask.
Delivered documentation and presentations to technical and non-technical stakeholders.

Education

Master of Science - Data Science

Digital University Kerala ( Former IIITMK )

Trivandrum, India

04.2001 -

Skills

Generative AI: GANs, VAEs, Diffusion Models, LLMs (GPT, BERT, LLaMA, Falcon)
Deep Learning: PyTorch, TensorFlow, Transformers, Transfer Learning
NLP: Hugging Face, Tokenization, Text Summarization, Embeddings
Data Science: Pandas, NumPy, Matplotlib, Scikit-learn

Programming: Python, Shell Scripting, SQL
Deployment: Flask, FastAPI, Docker, Streamlit, Google Colab, Google Cloud
Other Tools: Git, Jupyter, Weights & Biases, MLflow, LangChain
Soft Skills: Mentoring, Public Speaking, Content Delivery, Collaboration documentation, collaboration, and communication

PORTFOLIO PROJECTS

Project 1: Text-to-Image & Audio Generation Using GANs and Diffusion Models

Description:
Created a multi-modal GenAI model that generates both images and audio from text descriptions. Used Stable Diffusion for visuals and Tacotron2 + WaveGlow for speech synthesis.

Key Highlights:

Combined Diffusion Models + GANs for output generation
Fine-tuned model on Indian regional datasets for accents and style
Developed a Flask-based API to integrate with mobile/web apps
Explained the entire model architecture in internal workshops

Tools: PyTorch, Hugging Face Diffusers, Tacotron2, WaveGlow, Flask

Project 2: Personalized Learning Path Recommender using GPT and Embeddings

Description:
Built a system to generate individualized learning journeys based on user inputs (goals, time availability, current skill level) using GPT-4 + Embedding Similarity and prompt chaining.