Hi, I'm Abhay
AI Enthusiast
Computer Science | Artificial Intelligence | Data Science | Software Development | Research
About
My Introduction
Abhay is a graduate student at UC San Diego who is passionate about computer science and its applications in various domains.He has completed courses and certifications in Python, machine learning, and deep learning. He participated in several hackathons and competitions, winning the Nasscom Academic Grand Challenge in the Analytics domain for his innovative solution. He is always eager to learn new technologies and collaborate on projects based on deep learning.
Skills
My technical & miscellaneous skillsAI & Data Science
3+ Years XPComputer Vision
Natural Language Processing
Generative AI
Probability & Statistics
Data Analytics & Visualization
Frameworks & Libraries
Programming
3+ Years XPPython
C & C++
SQL
Computing
1+ Years XPAmazon Web Services
GPU & Distributed Computing
Misc
2+ Years XPGit
Linux
Raspberry Pi
Software Development
Leadership
Communication
Experience
My journey in Academics & Professional Internships
B.Tech - Computer Science and Engineering
SRM Institute of Science and Technology , KTR Campus CGPA : 9.56 (Final)As an undergraduate student researcher :
- Co-authored two impactful research papers on Gait Speed-Based Individual Recognition using thermal images and presented findings at IEEE Explore International Conferences.
- Published a study comparing DenseNet201 and MobileNetV2 for medical image analysis using a 2-D Convolutional Neural Network.
- Collaborated with Dr. Vijaya K and Mrs. Nithyakani P at SRM IST, bridging the gap between computer science and healthcare.
Class XII
Maths, Physics, Chemistry, Computer Science | ISC , IndiaClass X
ICSE , India
Research Intern (IRCC)
Indian Institute of Technology Bombay - Incorporated Computer Vision techniques for affective engagement to analyze peak emotions in 100+ participant interactions, prepared descriptive statistics on a study log dataset of 80,000+ entries, and applied open-source valence-arousal models.- Annotated over 30,000 video frames and utilized YOLOv8 for detecting cognitive engagement in a classroom environment using the ICAP framework, leveraging DGX machines for video analytics.
Intern Analyst (CEDA)
Wells Fargo - Developed a Python-based data extraction tool using regex, text parsing, and NLP, reducing manual Handbook processing time by 70% and improving structured data accuracy by 30%.- Applied Latent Dirichlet Allocation (LDA) topic modeling on FHA Mortgage policies, achieving a 20% increase in classification accuracy and a coherence score of 0.78.
Research Intern (PRISM)
Samsung R&D Institute India-Bangalore - Conducted forecasting on a French bakery dataset, achieving a 15% reduction in forecasting mean absolute errors using algorithms like SARIMA, Holt-Winters, LSTM with training windows from 30-180 days.- Executed time series analysis exploring mean and standard deviation trends, seasonal effects, and confirmed data stationarity with the Augmented Dickey-Fuller (ADF) test, with a p-value < 0.05 for 95% confidence.
Academic Intern
National University of Singapore - Leveraged Spacy for Named Entity Recognition and Deep neural network to develop a robust Resume Parser and multi-label classifier, achieving an impressive precision rate of 96% for the classification of developer resumes.- Pioneered integration of cutting-edge Deep Learning optimization techniques; achieved a 10% increase in classification precision and streamlined workflows, elevating team productivity by 15%.
Projects
My works, projects & contributions
Asset Bubble
Project helps the investors using a predictive analytics model or providing actionable analytical insights to this problem. The output is the confidence score on whether the stock market will crash in 2022 or not and if it does find an approximate time interval in which this can possibly occur.
View
Alveoli
Chest X-Ray reports in a matter of seconds.We have created a secure Chest X-Ray image classification based website and app that can detect Covid-19, Pneumonia and Tuberculosis. Deep Learning has been used to detect the disease by using a Convolutional Neural Network(MobileNetV2) which performs classification
In Dev
ResPas
This project is aimed at building a multi-label classification model using the principles of Deep Learning and NLP to effectively classify users into job categories given their resume.
View
BTC Prediction
In this Deep Learning Application, we have used the price data for Bitcoin to forecast its price in a specified future window. We have used the Tensorflow and Keras APIs to build a stacked LSTM model.
View
Astro Numbers
This project performs digit recognition using deep learning concepts. It can classify an image into 10 classes.We have built a Multilayer perceptron (MLP) model using most popular Google library Tensorflow to recognize handwritten digits.
View
License Plate
Number Plate recognition, also called License Plate recognition using image processing methods is a potential research area in smart cities and the Internet of Things. An exponential increase in the number of vehicles necessitates the use of automated systems to maintain vehicle information for various purposes.
In Dev
InfoViz
The dataset would also include other variables, such as unemployment rates, industrial production levels, and stock prices, depending on the specific economic indicators being analyzed. The visualizations cover a range of economic indicators, including time, unemployment data, consumer price index, price-to-earnings ratio, open, high, low, close, and industrial production index.
ViewPublications
My research & publicationsGait speed based individual Recognition model using deep 2-D convolutional neural network
Conference : International Conference on Computer Communication and Informatics (ICCCI) | Progress: PublishedLink: 10.1109/ICCCI56745.2023.10128342
Author(s): Abhay Lal; P Nithyakani
Developed a custom 2-D Convolutional Neural Network using thermal infrared night images from CASIA C dataset for gait-based individual recognition, comparison and optimized classification of walking conditions outperformed pre-trained models.
Human Gait Recognition Using Cross View Micro Gait Dataset with Light weight MobileNet
Conference : International Conference on Recent Advances in Electrical, Electronics, Ubiquitous Communication, and Computational Intelligence (RAEEUCCI)Link: 10.1109/RAEEUCCI57140.2023.10134510
Author(s): Haripreeth Dwarakanath Avarur; Abhay Lal; Nithyakani P; Aryan Sinha; Gajulapalli Naga Vyshnavi; Shruthi Kannan
Implemented gait recognition using a lightweight MobileNet model on the Custom-made Cross View Micro Gait Dataset, surpassing existing methods for human identification from cross-view videos.
ALATS: Analysis of localization algorithms in terrestrial surveillance bots
Conference : IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT)Link: 10.1109/CONECCT57959.2023.10234740
Author(s): Vaishnavi Moorthy; Darshil Shah; Shresth Kapoor; Srinivas Tb; Abhay Lal
Designed and executed a state-of-the-art surveillance system on a Raspberry Pi, harnessing the YOLOv7 model and OpenCV to craft a compact, high-performance autonomous surveillance robot.