About
I’m Keerthana S, a B.Tech. student in Mechanical Engineering at IIT Guwahati with a strong passion for AI and machine learning. Over the past few years, I’ve interned at institutions like Harvard University, IISc Bangalore, and IIST Trivandrum, where I worked on diverse projects including large-scale web scraping, coarticulation analysis using Euclidean distances, and building custom CNNs for facial emotion recognition.
I enjoy solving real-world problems with deep learning and have led projects like monitoring manufacturing processes and designing multi-intent transformer networks for recommendation systems. I'm proficient in Python, TensorFlow, PyTorch, and other data tools, and I’ve also been fortunate to receive recognitions such as the AWS AI/ML Scholarship and high ranks in multiple Kaggle competitions. I'm always eager to learn, build, and contribute to impactful solutions.
Work
Harvard University
|Data Engineer Intern
→
Highlights
Implemented web scraping algorithms like Automated and API based for IEEE and ACM conference papers
Scraped over 45,000+ research papers and calculated LLM tokens (200 million+) using OpenAI GPT-4 tokenizer.
The collected material is crucial to validate the model's accuracy using OpenAI Models(GPT-3.5, GPT-40, GPT-4)
Spire Lab, Indian Institute of Science
|AI/ML Intern
→
Highlights
Coarticulation on velar consonants /k/ and /g/in VCV sequence, analyzing 74 subjects x 7 frames = 3108 data.
The resulted (p-values nearly 0.0001) in the data analyze the symmetry among the frames in each subject.
Euclidean distance analysis shows that significant differences(p-value<0.05) occur at point -3 in the VCV sequence.
Indian Institute of Space-Science and Technology
|AI/ML Intern
→
Highlights
Developed a custom CNN architecture for emotion detection with 0.89% accuracy, surpassing pre-existing models.
With improved FER performance, able to address the limitations and provide a liable tool in current FER systems.
Resulted in a 4.5% increase in accuracy and enhanced performance in recognizing and classifying emotions.
Volunteer
Udgam, IIT Guwahati
|Manager
Highlights
Managed Campus Ambassador Program, a team of about 800+ student ambassadors to promote club's activities.
Education
Indian Institute of Technology, Guwahati
→
B.Tech
Mechanical Engineering
Grade: 6.76 (Current)
Kerala State Board
→
Senior Secondary
Grade: 98.08%
Kerala State Board
→
Secondary
Grade: 99%
Awards
AWS AI and ML Scholarship
Selected as one of the 1,000 students who applied for the program globally
Multi-Class Prediction of Obesity Risk Kaggle Competition
Secured a rank of 417 out of 3587 participants
LeetCode, Maximum Rating: 1593 | Weekly Contest 414
Secured a Rank of 3052 | Ranked in top 24.24%
Project at UIT Norway
Selected as one of the 50 students nationwide for this remote internship.
Playground Series S3E11 Kaggle Competition
Secured a rank of 69 out of 952 participants
All India Women Hackathon
Selected among 50 semi-finalist out of 1000+ participants
Skills
Programming
Python(NumPy, pandas, scikit-learn), C, CPP, HTML, CSS, Javascript, React, R.
ML Frameworks
Keras, TensorFlow, PyTorch, LLM, PySpark.
Miscellaneous
Git, MySQL, Power BI, GSSOC(open source), Cloud Management.
Key Courses Taken
Mathematics
Linear Algebra, Data Science Maths (Duke University).
Artificial Intelligence
Recommendation System Design Using Deep Learning, Computing with Signals.
MOOC's
Deep Learning Specialisation by DeepLearning. AI, AI programming with Python.