Projects / Research / Experience / Toolkit / Connect
I work at the intersection of data science, applied AI and engineering. My projects span battery-health prediction, computer vision, document processing and data visualisation, with experience in healthcare R&D, analytics and freelance ML development.
Based in Berlin, I am pursuing an M.Sc. in Data Science, AI & Digital Business and an M.Sc. in International Business Management at GISMA. My background is in Electronics & Telecommunication Engineering, and I enjoy connecting physical systems, data and business questions.
What interests me: battery and energy analytics · applied machine learning · business intelligence · process automation.
Explore a repository for its methods, documentation and project workflows.
|
EV lithium-ion battery SOH estimation Machine learning for battery State of Health, connecting charging data, feature engineering and gradient-boosted models.
Battery SOH modelling · research documentation Explore the research → |
FDA food recall analysis Explore recall patterns, classifications and reporting trends through question-led analysis and visual storytelling.
Historical data analysis · notebook and chart explanations Explore the analysis → |
|
Great Britain road accident severity analysis Exploratory analysis of recorded accidents from 2005–2015, examining severity alongside road, environmental and temporal conditions.
Descriptive associations in recorded accidents Explore the patterns → |
PCB component placement classification Transfer learning with MobileNetV2 to classify PCB component placement images and examine classification errors.
Prototype · transfer learning and classification-error analysis Explore the model → |
|
NYC taxi trip duration prediction Trip-duration regression using temporal and geospatial features, including distance, bearing and location clusters.
Notebook study · feature engineering and model comparison Explore the workflow → |
PDF context compression A document-processing approach combining PDF-to-Markdown conversion, sentence selection and BART summarisation for shorter context.
Evaluation in progress · documentation available Explore the evaluation plan → |
How these projects connect
- Engineering data → modelling: battery health and taxi duration connect feature design with prediction.
- Images → classification: PCB inspection explores transfer learning and classification-error analysis.
- Records → explanations: FDA recalls and road accidents focus on readable analysis and careful interpretation.
- Documents → shorter context: PDF compression explores the balance between brevity and information retention; evaluation is ongoing.
Project scope and methods
EV lithium-ion battery SOH estimation — Documented battery State of Health modelling through charging-data preparation, feature engineering and LightGBM/XGBoost research.
FDA food recall analysis — Analysed historical FDA recall records using pandas and Plotly, presenting recall classifications, reporting trends and geographic patterns with chart explanations.
Great Britain road accident severity analysis — Explored Great Britain accident records from 2005–2015 through data cleaning, severity comparisons and visual analysis of road, environmental and temporal conditions.
PCB component placement classification — Developed a PCB placement classification prototype through image preprocessing, MobileNetV2 transfer learning, model tuning and classification-error analysis.
NYC taxi trip duration prediction — Built and compared trip-duration regression models using temporal and geospatial features, including distance, bearing and location clusters.
PDF context compression — Documented a PDF-to-Markdown, sentence-selection and BART summarisation workflow with an evaluation plan for context reduction and information retention. Evaluation is in progress.
My coauthored work connects machine learning, embedded systems and energy applications.
| Publication | Focus | Read |
|---|---|---|
| EV Lithium-Ion Battery SOH Estimation Based on LightGBM Model · Springer, 2025 | Battery-health estimation from charging data using LightGBM | Paper ↗ |
| The Fitness Tracker Band · AIP, 2025 | Embedded activity tracking with an ESP controller and Firebase integration | Paper ↗ |
| Sun Flora: Dual Axis Solar Tracker using 8051 in Assembly Language · IRJMETS, 2023 | Light-sensing, motor control and an assembly-language tracking system | Paper ↗ |
I worked on problem analysis, project planning, hardware selection and the device's algorithmic logic. I developed the ESP-based fitness-band device and integrated the hardware with Firebase for real-time data transmission. The project was a team effort, with collaborators contributing the Flutter app, enclosure, data collection, testing and documentation.
A System and Method for Calculation of State of Health of Li-Ion Battery
South Africa · Patent 2024/03681 · 2024
B.Tech project outcome in battery-health diagnostics.
Huawei TechArena 2026 — Nuremberg, Digital Power / CellScale
Worked on LFP cell State of Health trajectories and end-of-life prediction from cycling time-series data; qualified for the on-site ceremony.
Huawei TechArena 2026 — Munich, 3D-Matters, team entry
Worked on anatomical ear-landmark extraction from 3D head scans using a DiffusionNet ensemble and anatomical post-processing.
- Aavishkar 2023–24: qualified for the SPPU University Research Competition.
- PROTECH 2020: first place, project competition, Symbiosis Institute of Technology.
- Persona Fest 2020: second place, Glider Design & RC Aircraft Flying, MIT Pune.
Philips India Limited · Apprentice / Intern, Image Guided Therapy — Apr 2024–Jun 2025
Healthcare R&D · Pune · full-time, on-site
- Developed and validated a battery-health diagnostic proof of concept for Energy Storage Units.
- Built a prototype data-logging and analysis tool and automated manual diagnostic steps.
- Supported production-fault root-cause analysis and shared findings with R&D and production teams.
Nitya Innovations LLP · Data Analyst Intern — Jan–Jun 2024
Pune · remote, project-based
- Developed an EV battery State of Health prediction workflow with data cleaning and feature engineering.
- Trained and tuned XGBoost and LightGBM models in Python.
- Contributed research and methodological documentation on lithium-ion battery analysis.
Bunnyfied Labs · Jr. Data Analyst & Machine Learning Developer — Apr 2023–Jan 2024
Mumbai · freelance, part-time, remote
Helped build a resume-to-job matching prototype using TF-IDF and cosine similarity, contributed to Power BI prototype dashboards for booking, user and payment KPIs, and supported client-data cleaning and recurring-report automation with Python.
| Area | Tools and methods |
|---|---|
| Data analysis | Python · pandas · NumPy · SQL · Excel |
| Machine learning | scikit-learn · XGBoost · LightGBM · feature engineering |
| Deep learning & NLP | PyTorch · TensorFlow/Keras · Hugging Face · BART · TF-IDF |
| Visualisation & BI | Plotly · Matplotlib · Power BI |
| Engineering & embedded systems | MATLAB · ESP controllers · Firebase · 8051 assembly · Battery health analytics |
Education
- GISMA University of Applied Sciences, Berlin — M.Sc. Data Science, AI & Digital Business + M.Sc. International Business Management, 2025–2027 (in progress).
- Vishwakarma Institute of Information Technology, Pune — B.Tech Electronics & Telecommunication Engineering, 2021–2024 · CGPA 9.19.
- Cusrow Wadia Institute of Technology, Pune — Diploma Electronics & Telecommunication Engineering, 2019–2021 · 96.94%.
Interested in applied ML, battery analytics, research or data storytelling? Let's connect.