About Me
End-to-end AI/ML builder with a strong foundation in physics, hands-on IoT and embedded systems experience, and demonstrated ability to deploy machine learning solutions in real-world, resource-constrained environments. …
End-to-end AI/ML builder with a strong foundation in physics, hands-on IoT and embedded systems experience, and demonstrated ability to deploy machine learning solutions in real-world, resource-constrained environments. Experienced in collaborating with research institutions and domain experts to translate complex problems into practical, impactful AI solutions. Passionate about responsible AI for public good in LMICs.
Experience
AI/ML & Data Consultant (Freelance)
Collaborated with Bharathidasan University to develop an ANN-based prediction interface estimating optimal algae dosage for polluted water treatment
Reduced calculation time from several days to seconds with 95%+ model accuracy
Deployed on AWS SageMaker and EC2
Designed and deployed an automated plant growth monitoring system for a PhD candidate using Raspberry Pi and Arduino
Engineered UV day/night simulation, dual-camera image capture every 30 minutes, and real-time temperature/humidity logging
System handed over and still operational
Developed an R programming package for research purposes
Currently in the process of submission to CRAN
Built automatic data logging systems using Arduino for various research and IoT use cases
AI/ML & Data Consultant (Freelance)
Collaborated with Bharathidasan University to develop an ANN-based prediction interface estimating optimal algae dosage for polluted water treatment — reduced calculation time from several days to seconds with 95%+ model accuracy; deployed on AWS SageMaker and EC2, Designed and deployed an automated plant growth monitoring system for a PhD candidate using Raspberry Pi and Arduino — engineered UV day/night simulation, dual-camera image capture every 30 minutes, and real-time temperature/humidity logging; system handed over and still operational, Developed an R programming package for research purposes — currently in the process of submission to CRAN, Built automatic data logging systems using Arduino for various research and IoT use cases
Data Analyst
Designed and maintained automated Python-based ETL pipelines to ingest, clean, and deliver sanitised data into Power BI dashboards, enabling weekly C-suite business insights
Reduced pipeline execution time from 40 minutes to 9 minutes (77% improvement) by redesigning input file formats
Built an end-to-end daily automation workflow ingesting customer data from the company website, processing via cleaning scripts, pushing to SharePoint, and auto-refreshing Power BI
Collaborated with manager to identify manual workflows and convert them into automated Python solutions, reducing operational overhead across the team
Data Analyst
Designed and maintained automated Python-based ETL pipelines to ingest, clean, and deliver sanitised data into Power BI dashboards, enabling weekly C-suite business insights, Reduced pipeline execution time from 40 minutes to 9 minutes (77% improvement) by redesigning input file formats — a systematic, root-cause approach to performance optimisation, Built an end-to-end daily automation workflow ingesting customer data from the company website, processing via cleaning scripts, pushing to SharePoint, and auto-refreshing Power BI — completed within a 2-hour window, 6 days a week, Collaborated with manager to identify manual workflows and convert them into automated Python solutions, reducing operational overhead across the team
Data Science Intern
Applied ML and Deep Learning techniques to real-world client projects under mentorship
Gained hands-on experience in model development, evaluation, and deployment workflows
Data Science Intern
Applied ML and Deep Learning techniques to real-world client projects under mentorship, Gained hands-on experience in model development, evaluation, and deployment workflows
PROJECTS
Audio Frequency Image Processing
Converted Audio signals into mel-spectrogram images, created data pipelines and trained CNN and Transformer models to classify the genres. Deployes the model in Hugging Face for user interaction.