Arjun Selvakumar
Turning messy data into models and dashboards people can actually act on — and shipping AI products end to end, from real-time voice agents to interoperable multi-agent systems.
- Roles held
- 0
- Projects built
- 0
- Tools & skills
- 0
About Me
Background
Data science student with a foundation in AI engineering and a bias toward work that ships.
M.Sc. Data Science student at Friedrich-Alexander-Universität Erlangen–Nürnberg (FAU), building on a B.Tech in AI & Data Science from Rajalakshmi Institute of Technology, Chennai. Focused on applied AI systems and data-driven problem solving.
Hands-on experience across the full data science pipeline — data cleaning, exploratory analysis, statistical modeling, and predictive systems. Currently exploring AI agents and LLM-driven applications. Working knowledge of production machine learning workflows and real-world deployment challenges.
Primarily Python and TypeScript. Comfortable with SQL, PyTorch, and business intelligence tools. Building full-stack projects that combine backend AI systems with frontend web interfaces.
Languages
- GermanB1
3 / 6 on the CEFR scale
- EnglishIELTS 7
7 / 9 on the IELTS scale
Certifications
- 11/2024 – 12/2024
Power BI
Simplilearn
- 07/2022 – 08/2022
Machine Learning
Sololearn
- 04/2022 – 05/2022
Master Course For Python Programming
Pantech E-Learning Pvt. Ltd
Without data, you're just another person with an opinion.
Career
Experience
Hands-on data science work across industry and university research labs.
AI Intern
10/2026 – PresentIndiekidz GmbH
- Developing AI-powered applications and agentic workflows using LangChain and LangGraph, with a focus on building modular, reliable and maintainable LLM-driven systems.
- Building modular LLM orchestration pipelines that connect language models, tools and application logic.
- Implementing state-based workflows, conditional routing and agent execution.
- Integrating, testing and improving AI components within application workflows.
- Contributing to practical AI solutions with an emphasis on reliability and maintainability.
Data Science Intern
08/2025 – 01/2026DLK Software Solutions
- Contributed to end-to-end Data Science and Machine Learning projects, including data cleaning, data preprocessing, and Exploratory Data Analysis (EDA).
- Performed data analysis and statistical modeling using Python, Pandas, NumPy, and Matplotlib to identify patterns, trends, and actionable insights.
- Developed and deployed predictive models for classification and regression using supervised Machine Learning algorithms.
- Conducted model validation, performance evaluation, and optimization using metrics such as Accuracy, Precision, Recall, and F1-score.
- Delivered insights through data visualization, analytical reporting, and stakeholder communication, supporting data-driven business decisions.
Work-Study – Data Science Lab
12/2023 – 05/2024Rajalakshmi Institute Technology
- Executed end-to-end Data Analytics and Predictive Modeling projects, including data preprocessing, feature engineering, and EDA using Python.
- Designed and developed interactive Business Intelligence (BI) dashboards using Power BI, Qlik Sense, and Tableau to monitor KPIs, trends, and performance metrics.
- Implemented ETL pipelines, data transformation, data modeling, and DAX calculations for multi-source data integration and reporting automation.
- Applied Visual Analytics, dashboard optimization, and structured reporting to enhance decision support systems and business intelligence processes.
Selected Work
Projects
A curated selection of work across AI, data science, and full-stack development.
PMW – Product Minds on Wheels
- Built during the EHL Grand Finale 2026 BMW challenge hackathon to help product managers make evidence-based product decisions.
- Integrates customer feedback, survey responses, sales figures and equipment data to identify important insights and support feature prioritisation.
- Includes a conversational voice assistant that allows product managers to explore feedback, understand prioritisation scores, and request recommended next steps.
- Implements a human-in-the-loop approach where AI prepares and explains recommendations, but a product manager reviews evidence and approves decisions before changes are made.
- Maintains an audit trail of all decisions to support transparency and accountability in product management workflows.
int_review – Voice Mock Interview & Salary Negotiation
- Built an end-to-end web app where you paste a resume and a job description, then a voice persona interviews you, asks three questions, and speaks a verdict out loud.
- Ran the voice loop on ElevenLabs Conversational AI — an Interviewer agent and an HR Negotiator agent connected over WebRTC, with resume, job description, persona brief, and salary band injected per session as dynamic variables.
- Used Claude Sonnet 5 for the judgment calls: one pass finds the resume's gaps against the job description before the call, another turns each transcript into a debrief covering STAR structure, the answer that decided the verdict, and a rewrite of the weakest answer.
- Measured filler words, hedging, repeated sentence starters, and answer-length outliers deterministically in code rather than asking a model for the numbers.
- Added a negotiation round where the HR persona holds a hidden salary ceiling, so the debrief can reveal the ceiling and how much of the range you actually captured.
A2A – Agent2Agent Protocol Implementation
- Implemented Google's open Agent2Agent (A2A) protocol in Python to enable interoperable communication between independent AI agents.
- Built Agent Cards for capability discovery, allowing agents to advertise their skills and negotiate tasks with one another.
- Designed task delegation and messaging flows over JSON-RPC/HTTP so agents built on different frameworks could collaborate on multi-step tasks.
- Explored practical multi-agent coordination patterns, laying the groundwork for interoperable, framework-agnostic agentic systems.
Healthcare Operations Dashboard – Power BI
- Designed an interactive Power BI dashboard to analyze patient waiting lists and healthcare KPIs.
- Implemented dynamic filters and slicers (time period, case type, specialty) for targeted analysis and self-service reporting.
- Visualized average & median waiting times, YoY comparisons, trend analysis, and ranking of specialties.
- Enabled data-driven decision-making through structured Healthcare Analytics and KPI reporting.
04 / Toolbox
Skills & Technologies
Languages, frameworks, and tools I use to build, analyze, and deploy solutions.
AI & LLM Applications
Building intelligent systems and agentic workflows
Programming Languages
Core languages and runtimes
AI & Web Development
Frontend frameworks and API integration
Data Analysis & Visualization
Data science and business intelligence tools
Big Data Tools
Distributed systems and data processing
MS Office Suite
Productivity and office applications
Soft Skills
Collaboration and professional development
Academics
Education
From an AI & Data Science bachelor's in Chennai to a data science master's in Germany.
Master of Science (M.Sc.) in Data Science
Friedrich-Alexander-Universität Erlangen–Nürnberg (FAU)
Grade: 2.7 (German grading system)
Bachelor of Technology (B.Tech) in Artificial Intelligence & Data Science
Rajalakshmi Institute of Technology
German Grade Equivalent: 2.15 · CGPA: 8.15 / 10
Diploma in Python Programming
Computer Software College
Grade: A
Get In Touch
Contact
Open to conversations about data science roles, theses, and collaborations.