Krishna Kankipati
A Bit About Me
🤖🌍 Introducing your friendly neighborhood Machine Learning Enthusiast and AI Wizard! Fueled by a lifelong fascination with robots and artificial intelligence, I've embarked on a thrilling journey to explore the magical realms of machine learning.
🔮 With my enchanting arsenal of skills in Statistical Data Analysis, TensorFlow, Machine Learning, and Deep Learning, I'm on a mission to cast spells of ML applications onto real-world problems and bewitching computer vision projects. Together, let's create a world where technology and humanity harmoniously coexist.
🧙♂️ My adventures include a mystical internship as an AI & Web Engineer, where I conjured up a multilingual chatbot to charm hungry patrons and boost restaurant businesses. As an AI Research Apprentice, I crafted an LSTM model using the enchanted TensorFlow framework, foreseeing adverse digressions in critical industrial processes with a remarkable 90% accuracy.
Projects
Generative AI Applications
Automatic Tweets Extraction and Analysis (MLOps)
Titanic Passengers Survival Prediction (MLOps)
NY Taxi data analysis and Taxi Fare Prediction (Data Engineering)
Classification of Rice crops from Sentinel-1 and Sentinel-2 data (MLOps)
• Q&A Generating Agent (OpenAI, Streamlit, GCP) - Q&A Agent
• Improved tweets extraction time by implementing an automated pipeline using Prefect (Flows, Deployments & Schedules), and TensorFlow for categorization, and created an analytical dashboard using D3.js and Streamlit. Leveraged AWS (Sagemaker, Lambda, EventBridge & S3) to shift the code package into the cloud.
• Developed a production-ready model package (ongoing) using Feature Engine, Scikit-Learn, Tox, Pydantic, PyTest, and FastAPI for model serving. Continuous testing and input validation improved model robustness and reliability. Achieved an accuracy of 85% on the test data.
• Built an end-to-end ETL pipeline using PSQL, Docker, Terraform, BigQuery, dbt, and Spark to analyze and predict taxi fares. Created a robust Machine Learning algorithm using hyperparameter tuning and preprocessing techniques. Achieved 98% accuracy in fare prediction and reduced data processing time by 70%.
• Developed a production-ready code package (ongoing) using Feature Engine, Scikit-Learn, Pydantic, PyTest, and FastAPI for model serving. Used TensorFlow’s high-level API for efficient model development. Achieved an accuracy of 90% on the test data and reduced the model training time by 40% using GPU acceleration.
Work Experience
Sept 2022 - December 2022
Sept 2022 - December 2022
August 2020 - October 2020
July 2020 - October 2020
February 2020 - March 2020
January 2020 - March 2020
AI & Web Engineer Intern - NolymitAI
Designed and developed a highly effective multilingual chatbot for the restaurant business, resulting in a 35% increase in orders and a 25% reduction in customer service requests and promoting customer satisfaction scores.
Prior to that, created a web application platform for small-scale businesses, integrating Machine learning applications that increased sales by 30%. Implemented Firebase, Node.js, and Explicit Content Detection features in the web application, resulting in a 20% decrease in inappropriate content and an increased customer retention rate.
Streamlined project management by coordinating weekly meetings, preparing minutes, generating work plans, and producing work reports, reducing project completion time by 30% and increasing team productivity by 20%.
Graduate Student Assistant - Stevens Institute of Technology
Helped 50+ students understand challenging concepts and coordinated meetings with the professor when required for the graduate level Algorithms course.
Designed and evaluated assignment materials and midterm examination papers.
AI Research Intern - Widhya
Designed and implemented an LSTM model using the TensorFlow framework to predict adverse digressions in a steel manufacturing company's critical and industrial noise process with 90% accuracy, resulting in a 75% reduction in
unplanned downtime and a 60% decrease in maintenance costs and a 50% increase in production throughput.
Observed more than 95% change in system entropy when the machine approached a failure and formulated a
method to calculate the confidence score. This highly improved the reliability of the model's prediction.
By predicting failures early, reduced time delay in repairing and replacing machines, resulting in a 20% increase in production capacity, and a 50% reduction in equipment replacement costs.
Machine Learning Intern - Robofied
Produced a series of highly informative and comprehensive technical articles on Machine Learning and Deep Learning, and referencing research papers and technical reports to gain extensive theoretical knowledge and insight
Demonstrated technical proficiency by implementing the respective work with the Scikit-learn and TensorFlow framework, resulting in a 30% increase in website traffic and a 15% increase in engagement with the articles.
Machine Learning Intern - Cognia Technologies
Leveraged XGBoost Regression algorithm for predicting taxi demand in Hyderabad city.
Achieved a 90% explained variance score, which helped the client to monitor the taxi allocation effectively.
Machine Learning Intern - Exposys Data Labs
Reproduced a research paper on detecting DDoS attacks using a multi-classifier approach. Achieved 85% accuracy in detecting a DDoS attack.
Built an offline data model by ensembling different classification algorithms Logistic Regression, Random Forest, XGBoost, and Neural Network and used the majority voting method to finalize the prediction.
Optimized the model by creating a pipeline for data preprocessing, feature engineering, and model predictions.