ML Engineer | Full Stack Developer
ChaithuTalasila
Portfolio
I am a College Freshman studying at The University of Texas at Austin with a passion for computer science, Machine Learning, and Robotics. I have significant leadership and problem-solving experience. I hope to gain expertise in computer science and AIML through internships at STEM Companies.
I have extensive experience with competition robotics. I've been participating in the FIRST Suite of robotics competitions for 4 years as a Lead Programmer and Designer. Currently, I am the Lead Programmer on FIRST Robotics Competition Team 4192, where I designed motion profiling and planning software for holonomic robots and integrated AI-based computer vision based on TensorFlow and PyTorch image detection algorithms.
In recent years, I've developed an interest in AI and Machine Learning. When the pandemic started, I used lockdown as an opportunity to gain experience in this field by participating in the AWS DeepRacer competition and completing related projects. In 2021, I joined AWS DeepRacer Team as a team lead at Flower Mound High School to introduce and expose students to Machine Learning and AI and inspire the community to pursue STEM. In 2022, I became my school's team captain for AWS Deepracer and continue to further inspire my community to pursue STEM. I also took time to gain experience with OpenAI Codex and the GPT-3 Natural Language Processing engine. Currently, I am pursuing a Nano-degree in AI Programming with Python.
A python program I made during my time interning at Cisco that filters through CT scans of brains with potential tumors. The program then identifies critical images where the brains show signs of a tumor, identifies the location of the tumor, and identifies the percent accuracy of the program. This program helps automate and assist doctors with diagnosis, which is especially helpful with the rise of telehealth.
A python program that allows the user to easily customize and build their own custom AI model trained on Stable Baselines3 DQN and PPO algorithms to play Google Chrome’s dinosaur game. Models are assigned a random name, compressed, and saved to the desired location of the user and can be reimported easily to play the game again.
A model trained on Sklearn’s RandomForestClassifier with over a hundred thousand credit card transactions. The pipeline is built in such a way that anyone can switch out the data easily and the model will function just as well, if not better. Without tuning any hyperparameters and only using an 80-20 split between training and testing data respectively, I was able to achieve 100% accuracy on only my second try. This project is open source and can be found on my Kaggle.
Place where I post my career updates and can be contacted by anyone for a job or help with anything related to ML or Full-Stack Development.
The place where I post most of my projects (including this website), the majority of which are open source under the MIT license and can be downloaded and used by anyone.
Place where I post the code for some of my machine learning projects, such as the credit card fraud detector. I usually don't post on Kaggle as much as Github and mainly use it for its Jupyter Notebook and ability to quickly import large datasets.