Currently studying at KIT

HarshulNanda

>_ M.Sc. Computer Science @ KIT

I build practical, data-driven systems. My work ranges from automated measurement pipelines to reinforcement learning agents for energy market simulations.

Résumé
Karlsruhe, Germany

An orbit of how I work.

I work across several connected fields. Explore the orbit to see what I build and how the pieces fit together.

Explore the orbit

Move over a node or tap it to learn more. Select the centre to see who I am.

Where I've worked.

My work has taken me from battery analytics and software engineering to reinforcement learning research at KIT.

Jan 2026 to present Karlsruhe, GermanyOn-site

Student Research Assistant

Karlsruhe Institute of Technology

Researching reinforcement learning for energy market simulations within the open source ASSUME framework.

  • Developing and integrating a Multi-Agent Proximal Policy Optimization algorithm in ASSUME to model complex energy market behaviour.
  • Developing and integrating parameter sharing functionality for reinforcement learning algorithms within the ASSUME framework.
Jun 2024 to Jul 2025 TurkeyRemote

Battery Data Engineer

MEAtec

Developed machine learning pipelines for industrial battery analytics and enabled a small team to deliver models consistently.

  • Built Python pipelines for cell level SOC, SOH, RUL and anomaly detection. Shipped each model as a reusable package with automated GitHub Actions workflows.
  • Produced synthetic battery data with Equivalent Circuit Model simulations and built the supporting cleaning, feature engineering and validation steps.
  • Designed a Streamlit dashboard for responsive cell metric analysis and documented the full system with clear UML diagrams.
  • Introduced shared templates for documentation, reports and GitHub packages to keep the growing team's work traceable.
Jun 2023 to Aug 2023 Noida, IndiaOn-site

Machine Learning Intern

Havells Pvt. Ltd.

Forecast sales across several product lines using statistical models and neural networks.

  • Applied time-series forecasting with 10+ statistical models (ARIMA family) and LSTM neural networks to predict future sales across diverse product lines.
  • Contributed to a 45%+ improvement in error rate compared to traditional predictive models.
Jan 2023 to May 2023 IndiaRemote

Software Engineer

Paras Labs

Developed a secure and approachable iOS banking application.

  • Helped develop a secure digital-banking iOS application using UIKit and SwiftUI.
  • Built user-friendly interfaces with intricate designs and logic.
Jun 2022 to Nov 2022 IndiaRemote

Machine Learning Intern

BeyondExams

Collected a large YouTube dataset and trained a classifier for educational content.

  • Scraped 10,000+ YouTube videos into channel-specific and video-specific intensive datasets.
  • Built an innovative model architecture for categorical classification of YouTube videos (educational vs. non-educational).
Apr 2022 to Jun 2022 Tokyo (Remote)Remote

Computer Science Technical Writer

OpenGenus

Wrote practical, detailed articles about algorithms and data structures.

  • Published articles explaining algorithms and data structures with practical implementation, diverse methods and mathematical rigor.

Research & projects.

A selection of projects in embedded systems, autonomous robotics, data engineering and applied machine learning.

01
NLPRoBERTaRHugging FaceMarketing Analytics

May 2026

Sentiment Analysis for Digital Marketing

A RoBERTa sentiment pipeline that examines Ryanair Flex Plus customer comments.

  • Built an R-based sentiment-analysis pipeline for Ryanair Flex Plus customer comments using a RoBERTa transformer model through Python integration.
  • Conducted visual analysis across sentiment-distribution charts, word clouds, n-gram analysis, engagement comparisons and aspect-based sentiment trends.
02
Embedded SystemsArduinoBLESensor FusionIoT

Jan 2026 to Feb 2026

Smart Intrusion Detection with BLE

A portable intrusion detector with password protection and multi-factor BLE verification.

  • Designed a portable, password-protected intrusion-detection system on a Seeed XIAO nRF52840, fusing an IR flame sensor (light-change detection) with an LSM6DS3 IMU (tamper / movement detection).
  • Combined BLE proximity, live authentication through custom GATT characteristics and a dynamic password with a secure reset flow.
  • Built a firmware architecture with distinct audible feedback patterns for authentication, intrusion and error states.
03
ROSTurtleBot-3ArduinoRaspberry PiControl Theory

Dec 2024 to Jun 2025

Autonomous Trajectory Tracking in Mobile Robotics

Fractional PID control and obstacle avoidance on a TurtleBot 3, tested in simulation and on real hardware.

  • Compared standard, fractional, adaptive fractional and nonlinear fractional PID controllers for trajectory tracking.
  • Proposed a general PID control mechanism, validated both in simulation and on a physical TurtleBot-3.
  • Implemented and improved four pathfinding algorithms (PSO-Heuristic A*, Dijkstra, BFS, Greedy Best-First), integrating them with PID for static obstacle avoidance.
04
Web DevelopmentDashboardsRandom ForestsGaussian Processes

Nov 2024 to Dec 2024

Predictive Health Management System

Patient & admin dashboards predicting hospital-readmission risk in real time.

  • Developed integrated patient and admin dashboards for real-time readmission-risk tracking.
  • Implemented predictive models (Random Forests, Gaussian Process Classification) with an easy re-training platform for adaptive risk estimation.
05
94%
classification accuracy
Video ClassificationDeep LearningOpenCV

Nov 2023 to Dec 2023

Predicting Unethical Activity in Videos

Deep-learning video classification separating criminal from safe content.

  • Curated a dataset of CCTV footage and videos containing unethical activity, then analysed four video-classification model architectures.
  • Achieved over 94% accuracy classifying criminal vs. safe content.

Published in the Springer proceedings of the International Conference on Generative AI, Cryptography and Predictive Analytics.

06
98%
BERT accuracy
NLPBERTData Science

Feb 2023 to Apr 2023

Crime Prediction in Conversations

A fine-tuned BERT model flagging unethical intent across everyday dialogue.

  • Built a dataset of 1,000+ daily life conversations with 7 to 10 exchanges each, annotated across six categories from unethical to neutral.
  • Studied and compared 50+ crime-predicting text-classification architectures.
  • Proposed a state-of-the-art fine-tuned BERT architecture for public-safety NLP, reaching 98% accuracy on the created dataset.

Skills & technologies.

Move over a segment or its label to see the tools I use in that area.

PythonRustPyTorchReinforcement LearningTensorFlowPandasNumPyScikit-learnNLPOpenCVROSDockerCI/CDStreamlitFlaskReactJSLangChainMATLABBlenderPythonRustPyTorchReinforcement LearningTensorFlowPandasNumPyScikit-learnNLPOpenCVROSDockerCI/CDStreamlitFlaskReactJSLangChainMATLABBlender
37
skills
PythonRustC / C++JavaScriptMATLABHTMLCSSMySQLMachine LearningReinforcement LearningPyTorchTensorFlowKerasScikit-learnNLPARIMA / LSTMNumPyPandasOpenCVMatplotlibSeabornLangChainStreamlitFlaskReactJSFigmaSwiftUIUIKitGitCI/CD · GitHub ActionsDockerROSBlenderData Structures & AlgorithmsNumerical MethodsLinear AlgebraObject-Oriented Programming
CommunicationCritical ThinkingProblem-SolvingAttention to DetailAdaptabilityCollaborationLeadershipPublic SpeakingCreativityCommunicationCritical ThinkingProblem-SolvingAttention to DetailAdaptabilityCollaborationLeadershipPublic SpeakingCreativity

Education & achievements.

Ongoing

Oct 2025 to present

M.Sc. in Computer Science

Karlsruhe Institute of Technology

Karlsruhe, Germany

Nov 2021 to Jun 2025

B.Tech. in Information Technology & Mathematical Innovation

Cluster Innovation Centre, University of Delhi

Final grade: 9.35 CGPA, approximately 1.3 to 1.5 in the German system

Apr 2020 to Jun 2021

Higher Secondary (Class XII)

Adarsh Public School

Final grade: 91.2%

Certifications

Move over a medal to preview it. Select the arrow to open the credential.

Achievements & Activities

A few milestones from olympiads, workshops and a hackathon win.

OlympiadDec 2020

Indian Computing Olympiad, Zonal Round

IARCS

Competed in the Zonal Informatics Olympiad 2021 (Indian Association for Research in Computing Science).

OlympiadApr 2021

International Mathematics Olympiad

Science Olympiad Foundation

Participated in the International Mathematics Olympiad conducted by SOF.

1st PlaceNov 2022

Hackathon Winner at Think Beyond Exams

BeyondExams

Led my team to 1st place, turning educational-classification models into a web app that predicts a YouTube video or playlist's educational percentage from just a URL.

WorkshopNov 2023

Facilitator for the Faculty Development Program

Cluster Innovation Centre

Hosted a two-day Blender 3D workshop, animating scientific phenomena (electron transmission, wave propagation) to help teachers communicate complex science.

WorkshopAug 2025

EV Technology & Public Charging Station Workshop

NIELIT

A five day workshop on EV technology covering battery dynamics, motors, power electronics and charging.

Let’s build something worth shipping.

Open to research and engineering roles, collaborations and a good technical conversation. The fastest way to reach me is email.

© 2026 Harshul Nanda.