SHEET 01 / 07 — PROFILE
Chirayu Jaju
builds ML & deep learning systems.
B.Tech CSE & AI, VIT-AP University · Class of 2027
Machine Learning Engineer with hands-on experience designing, training, and deploying ML, deep learning, and generative AI models across predictive analytics, computer vision, and NLP — backed by MLOps practice with Docker, MLflow, and AWS.
About
SHEET 02 / 07I'm a final-year B.Tech Computer Science & Artificial Intelligence student at VIT-AP University, focused on machine learning, deep learning, and generative AI — with production-minded MLOps experience layered on top.
My work spans predictive analytics on tabular data, computer-vision and sequence models for activity and medical-image classification, and shipping those models behind FastAPI services in Docker containers on AWS.
I care about the full pipeline — clean data, sound evaluation, and a model that actually survives contact with production.
Skills
SHEET 03 / 07Experience
SHEET 04 / 07Built classification models (Logistic Regression, Random Forest, XGBoost) for diabetes prediction with over 90% accuracy through feature engineering and hyperparameter tuning. Built a CNN+LSTM model in TensorFlow/Keras for human action detection from video, improving accuracy by 15% via data augmentation. Trained time-series regression models for bike-sharing demand forecasting, reaching an R² of 0.92.
Developed and executed manual and automated test cases (unit, integration, regression) across the SDLC. Built automated test scripts in Python and Selenium, improving test coverage and cutting manual testing effort. Tracked defects and collaborated with developers on debugging, code reviews, and release validation.
Projects
SHEET 05 / 07Human Activity Recognition system on wearable sensor data (MHEALTH dataset) classifying walking, jogging, cycling, and climbing stairs. Compared Logistic Regression, KNN, SVM, Decision Tree, Naive Bayes, and Random Forest with robust preprocessing and cross-validation, reaching over 92% classification accuracy.
Hybrid CNN-LSTM deep learning model to detect and classify brain tumors from MRI scans, combining spatial feature extraction with sequential pattern learning. Image preprocessing and augmentation improved generalization on the MRI test set.
Classification pipeline comparing Logistic Regression, Random Forest, and XGBoost with careful feature engineering and hyperparameter tuning, reaching over 90% accuracy.
Deep Q-Learning based adaptive traffic signal control system using SUMO and TraCI. The agent learns traffic-light policies from an 80-dimensional traffic state to reduce vehicle waiting time and improve traffic flow at a 4-way intersection.
Interactive Blackjack application powered by Monte Carlo Reinforcement Learning. The agent learns a Q-table from simulated Blackjack hands and uses the learned policy to recommend actions while playing against the user in real time.
Time-series regression models forecasting bike-sharing demand, achieving an R² of 0.92 through feature engineering on temporal and weather data.
Credentials & Links
SHEET 06 / 07Certifications & Achievements
SHEET 07 / 07 — CONTACT
Let's build something that ships.
Open to ML engineering roles, internships, and collaborations — especially anything involving computer vision, NLP, or getting a model out of a notebook and into production.