Publications
Computational Humor Classification & Generation
Towards Conversational Humour Analysis and Design, Humor Research Conference 2021
- Comparative Analysis of Feature Based Machine Learning classifiers for detecting Humor using: Logistic Regression, Naive Bayes & Support Vector Classifiers.
- Achieved a F1 score for classification of 0.63 on the Reddit dataset.
- Explored Humor Generation using Bi-LSTMs in PyTorch.
Work Experience
Backend Developer, Cloudphysician
- Designed and set-up an intranet website for Cloudphysician from scratch using MERN stack.
- Created efficient database schemas & handled MongoDB clusters.
- Managed a team of 3 developers, oraganised client meets and scheduled deadlines, leading to a successful completion.
Hackathon Wins
Social Media Search Engine for Financial Markets
Echodex, Sentiment & Opinion Mining Natural Language API Hackathon, Expert.ai
- Aggregates data from Twitter, Reddit and News APIs, providing a comprehensive analysis of market sentiments and opinions.
- Handled and integrated multiple APIs from Expert.ai for the analysis.
Walmart Product Clustering using K-Means
Domain Specific Search Engine, TAMU Datathon 2020, sponsored by Walmart
- Trained an agent on Google’s Dialogflow platform, to extract semantic information from user queries.
- Leveraged TF-IDF with K-Means clustering to find Walmart products semantically related to extracted user inputs. Also recommend similar products.
Confession Generator
Confession Generator, IIIT-H Intra-Batch Hackathon
- Implemented and compared a character based LSTM model and GPT-2 model to generate text.
- GPT-2 was fine-tuned on undergrad college data.
[PoC] Chatbot from SMEs
Analysy.ai, Summer Up, E-Cell IIIT-H
- A Proof-of-Concept business administration tool for Small and Mid-Size Enterprises (SME).
- Worked on SpaCy pipelines for text-processing for the chatbot.
Projects I am Proud of
DistilBERT for Sentiment Analysis on Financial Data
- Fine-tuned the DistilBERT transformer model from Hugging-Face to extract Market Sentiments.
- Improved classification accuracy from 80% to 85%
Named Entity Recognition for FIFAWC 2018 Tweets
- Identifies named entities in twitter data for FIFA WC 2018 using Bi-LSTM Model with CRF, in Tensorflow 2.0.
- Gained accuracy of 93% on testing data.
Skills
- Agile
- C/C++
- Docker
- Git
- HTML
- JavaScript
- Keras
- MongoDB
- MySQL
- NLTK
- Pandas
- Python3
- PyTorch
- Shell Scripting
- Sklearn
- SpaCy
- Tensorflow 2.0
Education
B.Tech in Computer Science + M.S. by Research in Computational Linguistics (9.05 CGPA)
International Institute of Information Technology, Hyderabad
- Computer Programming (C)
- Computer Systems Organization (Assembly)
- Data Structures & Algorithms (C++)
- Algorithm Analysis & Design (Python3)
- Automata Theory (Python3)
- Introduction to Software Systems (HTML/CSS/JS)
- Design & Analysis of Software Systems (Agile + MERN)
Linguistic & NLP courses
- Intro to Linguistics I (Python3)
- Intro to Linguistics II (Python3)
- Computational Linguistics I (Python3 + NLTK + Tensorflow 2.0)
- Computational Linguistics II (Python3 + NLTK + PyTorch)
- Intro to NLP (Python3 + PyTorch)
12 Board CBSE - 93.4%
10th Board ICSE - 96.6%
Interests
- CP: LeetCode
- CyberSecurity: TryHackMe Profile
- Gymming
- Playing acoustic & bass guitar