I am a final-year Computer Science & Engineering (Data Science) student at Alva's Institute of Engineering and Technology (affiliated with VTU, Karnataka). I focus on building practical, production-ready machine learning systems, data pipelines, and scalable backend services.
My engineering philosophy is simple: machine learning is only as valuable as the software systems running it. I spend my time going beyond Jupyter notebooks — optimizing feature pipelines, tracking experiments with MLflow, wrapping models into sub-25ms asynchronous FastAPI microservices, containerizing with Docker, and deploying to AWS.
|
End-to-end supervised pipelines: cross-validation, hyperparameter tuning with Optuna, collinearity reduction, and tree ensembles (XGBoost, LightGBM, Random Forest) with SHAP interpretability. Python
Scikit-learn
XGBoost
Pandas
|
Acoustic speech transcription with OpenAI Whisper, automated PII sanitization with Microsoft Presidio, and low-latency conversational reasoning. OpenAI Whisper
Microsoft Presidio
NLP
|
|
On-device hardware inference on the NVIDIA Jetson Nano, continuous multi-sensor telemetry acquisition, and dynamic threshold automated solenoid actuation. NVIDIA Jetson Nano
IoT Telemetry
Edge ML
|
Asynchronous FastAPI microservices, MLflow experiment tracking & artifact registry, Docker containerization, and deployment on AWS EC2 & S3. FastAPI
MLflow
Docker
AWS (EC2/S3)
|
┌── [ 2023 — 2027 ] ─────────────────────────────────────────────────────────────────┐
│ B.E. in Computer Science & Engineering (Data Science) │
│ Alva's Institute of Engineering and Technology (AIET) │
│ Affiliated with Visvesvaraya Technological University (VTU), Karnataka │
└────────────────────────────────────────────────────────────────────────────────────┘
- Core Coursework: Machine Learning, Artificial Intelligence, Database Management Systems, Data Structures & Algorithms, Operating Systems, Cloud Computing, Object-Oriented Programming (C++/Python), Probability & Statistics.
- Engineering Focus: Designing reliable software architectures around machine learning models and data pipelines that solve real-world problems.
Categorized by technical domain:
Case studies demonstrating end-to-end architecture, mathematical modeling, and production deployments:
End-to-end tabular predictive pipeline, experiment tracking with MLflow, asynchronous FastAPI service, and containerized deployment on AWS.
- Data Preprocessing & EDA: Imputed, normalized, and transformed high-dimensional customer activity telemetry; handled categorical encoding and collinearity reduction.
- Model Training & Evaluation: Evaluated ensemble algorithms (XGBoost, Random Forest, Logistic Regression); optimized hyperparameters via stratified cross-validation, achieving 0.942 ROC-AUC.
- Experiment Tracking: Logged runs, evaluation metrics, and model artifacts with MLflow.
- FastAPI Microservice: Built an asynchronous FastAPI service for sub-25ms real-time churn risk inference.
- Docker & AWS Deployment: Containerized the entire inference runtime with Docker and deployed on AWS (EC2 & S3).
Python • Scikit-learn • XGBoost • MLflow • FastAPI • Docker • AWS EC2/S3
Speech-interactive AI platform with automated zero-knowledge PII sanitization and contextual sentiment analysis.
- Speech Processing: Integrates OpenAI Whisper for high-accuracy phonetic transcription from real-time microphone input.
- Privacy-Preserving PII Redaction: Integrates Microsoft Presidio to detect and anonymize personal identifiers (names, locations, contact info) prior to text processing.
- Empathetic Interaction Engine: Analyzes contextual sentiment and generates structured supportive responses in real time.
- Interactive Interface: Developed with Streamlit for responsive, cross-platform client interaction.
Python • OpenAI Whisper • Microsoft Presidio • Streamlit • NLP
IoT telemetry acquisition and edge machine learning system on NVIDIA Jetson Nano for automated precision irrigation.
- Edge Telemetry Processing: Deployed on the NVIDIA Jetson Nano platform to process multi-channel soil moisture, ambient humidity, and temperature telemetry.
- Dynamic Threshold Actuation: Implements dynamic threshold heuristics to actuate automated relay solenoid valves based on environmental conditions.
- Resource Optimization: Achieves up to 40% water savings while maintaining optimal soil hydration levels for agricultural yields.
NVIDIA Jetson Nano • IoT Sensors • Python • Edge AI • Hardware Relays
MERN stack productivity engine with multi-parameter query optimization, search indexing, and real-time analytics.
- RESTful Micro-Endpoints: Robust Express and Node.js API supporting complete lifecycle (CRUD) operations and schema-level validation.
- Search & Filter Optimization: Multi-criteria query filtering, text-based search indexing, and dynamic property sorting on MongoDB.
- Telemetry Dashboard: Dynamic velocity charts and visual completion metrics built with React state management.
MongoDB • Express.js • React.js • Node.js • REST APIs
CURRENTLY EXPLORING
├── → Advanced Machine Learning (Deep architectures, feature stores, automated feature selection)
├── → Data Science (High-dimensional statistical modeling & data analytics pipelines)
├── → Generative AI (Streaming LLM workflows & acoustic speech architectures)
├── → MLOps (Continuous training, model drift detection, automated CI/CD)
├── → Cloud Deployment (Containerized microservices & scalable deployments on AWS)
└── → Backend Engineering (High-throughput async APIs with FastAPI & caching)
|
|
|
|
|
|
Have an idea, opportunity, or interesting problem? Let's talk.
