Shreyas Bhat | Computer Science Enthusiast

Hi, I'm Shreyas Bhat.

A

A curious, fast-learning engineer with a strong mathematical foundation and a diverse technical skillset.

A short demo of an app I helped build

About

I am a Computer Science Grad Student at Columbia University. I enjoy problem-solving and coding. I am actively looking for SWE, Data Engineering or Data Science roles. I am passionate about developing complex applications that solve real-world problems impacting millions of users.

Languages: Python, C, C++, HTML/CSS
Databases: MySQL, Neo4j, MongoDB
Libraries: NumPy, Pandas, OpenCV, Tensorflow, MatPlotLib, Keras
Frameworks: Apache Spark, Docker, CodePipeline
Tools & Technologies:AWS: EC2, S3, Lex, Dynamo DB, Cognito, CloudWatch, CloudSearch, ElasticSearch

Actively looking for internships in software developement, data engineering, data science or machine learning.

Experience

Columbia University, CS Department

Teaching Assistant

Spring 2023: Teaching Assistant for Natural Language Processing with Prof. Daniel Bauer. Responsible for collaborating with a group of 10 TAs and conducting weekly office hours, creating and grading exams and home works.
Fall 2022: Teaching Assistant for Computational Linear Algebra with Prof. Daniel Hsu. Responsible for collaborating with a group of 10 TAs and conducting weekly office hours, creating and grading exams and home works.

Sep 2022 - May 2023 | New York, NY

Vivitsu Labs

Machine Learning Engineer Intern

Designed a self-learning agent to perform robotic tasks with reinforcement learning algorithms in python, conducted runs on 14 environments, reduced time complexity by 10% compared to Deep Q-learning algorithms
Integrated agent to OpenAI and Mujoco environments; conducted simulations on AWS servers and spearheaded weekly presentations and project update meetings
Received the PMRF (prime minister research fellowship - awarded to 500 best project proposals across the country by the government of India) in 2021 for advancement in AI technologies

Tools:

Jan 2021 - May 2021 | Bangalore, India

IIT Madras

Algorithms developer intern

Developed a parallelization algorithm to cover edges in bounded graphs enabling a significant improvement in time complexity while covering a higher percentage of edges compared to FPT algorithms for degree 4 graphs
Collaborated with a team of 5 and organized weekly paper reviews and presentations on project updates

Jan 2022 - May 2022 | Bangalore, India

Research Experience

Indian Institute of Science

Bachelor's Thesis

Analyzed the number of open clusters in bond percolation in d-dimensional integer space and the number of connected components in the continuum percolation model in d dimensional real space.
Derived quantitative bounds on rate of convergence of number of clusters in the central limit theorem using the Kolmogorov distance and the Kantorovich-Wasserstein distance

Jan 2022 - May 2022 | New York, NY

Projects

                
Recommendation Chatbot
                  Serverless, micro-service driven Chat Bot
                
AccomplishmentsTools: AWS, HTML, CSS, DynamoDB, ElasticSearch, Cloud Computing, REST API
Full-stack web application on AWS using Lex to recommend relevant restaurants to customers
                    based on previous orders, cuisine and time slots
Front-end in CSS and JavaScript, hosted on an S3 bucket, which can scale to thousands of users
DynamoDB to store data of restaurants scraped from Yelp API, parsed user queries using elasticSearch

Smart Photo Album
                  A smart photo album that recommends pictures by user request
                
AccomplishmentsTools: CI/CD, CodePipeline,Docker, HTML, CSS, JavaScript, DynamoDB
Developed a photo searching application leveraging AWS Services. Used Amazon Rekognition to label the image,
                    Amazon Lex to extract keywords and Amazon Transcribe to add voice-search capability
Built a frontend on S3 with AWS CloudFront enabled, ensuring scalability and HTTPS websockets support

BERT-Lexical Substitution
                  Finding the best word for lexical substitution using BERT and Wordnet
                
AccomplishmentsTools: Python, Tensorflow, Bert, NLTK, Wordnet, GCP
Used pre-trained BERT models to predict the best word to substitute in masked Language model tasks
Combined BERT and wordnet to achieve 80% accuracy on SemEval task

Shell in C
                  Wrote a Shell in C to excecute common commands
                
AccomplishmentsDesigned and implemented a shell in C to read and execute commands including change directory, help
                    and exit. Parsed input data using tokenization

Paragraph Classifier
                  Classifies text as high or low quality using Ngram language classifier
                
AccomplishmentsDeveloped a model to classify an essay as high or low quality using N-gram classification model; trained model on
                    parsed and pre-processed data from 5000 essays.