Machine Learning Data Science

Deepak Kandikattu

**************@*****.*** +1-720-***-**** LinkedIn Github

EDUCATION & CERTIFICATION

University of Colorado Boulder Aug 2023 – May 2025 Master of Science in Data Science CGPA: 3.97/4

Coursework: Information Visualization, Statistical methods, Data Mining, Machine Learning, Neural Networks, Text Mining. NVIDIA CERTIFIED GEN AI AND LLM ASSOCIATE [ LINK ] SKILLS

Languages: Python, R, SQL, HTML, CSS

ML & DS: Scikit-learn, XGBoost, TensorFlow, PyTorch, HuggingFace, CNN, RNN, LLM, RAG, NLP (spaCy, NLTK) Data Processing: Pandas, NumPy, Feature Engineering, Dimensionality Reduction, Data Cleaning, Pipelines (ETL, ELT) Visualization: Tableau, Power BI, Matplotlib, Seaborn, Looker Studio Cloud & Tools: GCP (BigQuery, GKE), AWS (S3, EC2), Google Cloud Storage, Docker, Git, Jupyter, Google Colab MLOps & Deployment: Flask, Kubernetes, Redis, Google Cloud Storage, CI/CD Basics EXPERIENCE

Junior Data Scientist, Fabhost Aug 2022 – Jul 2023

• Engineered a facial recognition-based attendance tracking system using Python, OpenCV, and scikit-learn, achieving over 95% detection accuracy and eliminating manual entry by digitizing attendance workflows across all departments.

• Implemented and validated a logistic regression model to predict employee retention risks from attendance and performance metrics, enabling the HR team to identify early warning signs and take data-driven actions.

• Built and deployed interactive dashboards in Tableau to visualize attendance trends, team productivity metrics, and arrival time patterns, improving visibility and enabling faster operational decisions.

• Constructed Python ETL pipelines to clean, normalize, and consolidate over 10,000 weekly records from multiple internal data sources, ensuring accurate, analysis-ready data for reporting and business intelligence use. Data Analyst Intern, Indian Space Research Organisation (ISRO) Dec 2019 – Jun 2020

• Collaborated with the MOTR (Multi-Object Tracking Radar) team to understand signal processing and refine algorithms that accurately track the position and path of satellites and launch vehicles using radar data.

• Analyzed real-time telemetry data and signal processing methods to maintain synchronized communication between ground control and onboard systems, which helped the team achieve a 100% mission success rate. PROJECTS

Interpretable Music Genre Classification using CNNs and Spectrogram Analysis Jan 2025 – May 2025

• Trained a 2D CNN in PyTorch for multi-class music genre classification using Mel-spectrograms from the FMA Small dataset, converted audio to .npy tensors, and performed hyperparameter tuning with cross-validation to improve model performance.

• Applied Grad-CAM to visualize which regions of the spectrograms the model focused on when making predictions, helping interpret the time-frequency features that distinguish different music genres. Wikipedia Articles Summarization and Conversion to Audio Aug 2024 – Dec 2024

• Engineered a Transformer-based system using Google Text-to-Speech to convert Wikipedia articles into audio, broadening accessibility for auditory learners and visually impaired users, and achieving 98% audio fidelity.

• Integrated Flask REST API, Google Cloud Kubernetes Engine, and Redis for caching and auto-scaling, enabling efficient handling of over 1,000 concurrent requests with seamless performance and zero downtime.

• Optimized data storage with Google Cloud Storage and BigQuery, ensuring reliable metadata management and analytics, delivering a robust platform for educational, research, and accessibility applications. Symptom-Based Disease Prediction using ML Classifiers and NLP Aug 2024 – Dec 2024

• Implemented a multi-class disease prediction system using Random Forest, Logistic Regression, and Naive Bayes, trained on 100+ symptoms across 1,000+ disease classes to provide real-time medical condition suggestions based on user-input.

• Achieved 92.31% accuracy using Random Forest on a structured medical dataset; built a user-friendly interface that expands input symptoms, suggests diagnoses, and integrates Wikipedia API for real-time medical summaries. Interactive Dashboard for Workplace Mental Health Analytics Aug 2023 – Dec 2023

• Developed an interactive dashboard using Tableau with 2,000+ data points from Kaggle to analyze the impact of mental health resources on productivity and stress across remote, hybrid, and onsite work environments.

• Generated visualizations to reveal key trends in productivity and stress levels across 3 different work environments.

• Discovered that remote workers with mental health support experienced a 15% boost in productivity, while hybrid and onsite workers exhibited a 20% higher likelihood of decreased productivity due to workplace stressors.

Contact this candidate