Vivek DevappallyShivanna
AI Engineer & ML Developer
Building intelligent systems with LLMs, RAG architectures, and agentic AI workflows. Specializing in autonomous AI agents, natural language processing, and production-ready ML solutions.
Hello!
Crafting autonomous AI agents and intelligent systems that solve real-world problems
Currently pursuing my Master's in Big Data and Artificial Intelligence at SRH Hochschule Berlin, I'm deeply focused on building production-ready AI systems using LLMs, RAG architectures, and agentic workflows.
As an AI Software Engineer at IAV GmbH Berlin, I develop AI-powered solutions including image understanding pipelines, intelligent data processing systems, and interactive visualizations using cutting-edge ML frameworks.
My AI projects range from autonomous market intelligence agents using LlamaIndex and GPT-4 to RAG-based document intelligence systems—always pushing the boundaries of what's possible with modern AI.
Years Experience
AI Projects Built
LLM Models Used
& Agentic AI
My Journey
Work Experience
September 2024 – Present
IAV GmbH
Berlin
AI Software Engineer
May 2022 – April 2024
Fledge Consulting
Bangalore
Associate Software Developer
March 2022 – May 2022
Fledge Consulting
Bangalore
Software Developer Intern
What I Do
Skills & Expertise
AI & Machine Learning
Programming & Frameworks
Tools & Infrastructure
Core Competencies
Featured Work
Recent Projects
Agentic AI & Financial Analysis
Stratagem AI – Autonomous Market Intelligence Engine
An autonomous AI agent that performs real-time financial research, market analysis, and competitor reconnaissance to deliver actionable investment verdicts.
- •Real-time web research with Tavily API bypassing knowledge-cutoff limitations
- •Live market data integration using yFinance for stock metrics and volatility analysis
- •Agentic workflow with multi-step reasoning using LlamaIndex
- •Custom Streamlit UI visualizing agent's thought process in real-time
AI & Machine Learning
RAG-Based Document Intelligence
Retrieval-Augmented Generation system that ingests user-uploaded files to provide accurate, context-aware answers and data extraction based on document content.
Academic Contributions
Research Papers
How Can AI Systems Be Made to Understand and Process Abstract Concepts
Alexander I. Iliev, Vivek Devappally Shivanna, Amirreza Soltani, Sanjeeth Chittyala
- →Led the experimental section by implementing five major embedding methods—t-SNE, PCA, LSA, Word2Vec, and Autoencoders—to model abstract concepts such as Freedom, Justice, Love, Happiness, and Knowledge.
- →Carried out full text-preprocessing using TF-IDF, designed numerical representations, and developed pipelines for dimensionality reduction and embedding generation.
- →Applied K-Means clustering and computed silhouette scores to quantitatively evaluate the semantic separability of different embedding models.
- →Built multiple 3D visualizations to compare embedding behaviours (cluster compactness, inter-cluster spacing, nonlinear structure retention, semantic preservation).
- →Contributed to analyzing the role of neural-symbolic reasoning, predictive coding, information geometry, and hypervector embeddings in improving conceptual reasoning in AI.
Design and Implementation of Smart Solar Panel Cleaning System
Vivek DS , Vaishali NB, V N Suraj Kumar, Devaraju D, Akshay Kumar R,
- →Developed and programmed the Arduino Uno firmware that controls the entire cleaning mechanism.
- →Implemented motor timing logic based on system calculations (rpm → linear distance → cleaning time).
- →Integrated the Arduino with the L298N motor driver to automate forward/reverse movement for complete panel sweeping.
- →Built logic for remote control of the system via a mobile app interface.
- →Participated in prototyping, testing, and refining the single-motor Locher rack mechanism for efficient dry cleaning.
An Efficient Automatic Solar Panel Cleaning System for Roof-top Solar PV System
Vivek DS , Devaraju D, V N Suraj Kumar, Akshay Kumar R, Galiveeti Hemakumar Reddy
- →Programmed the Arduino Uno for full automation of the cleaning system, including time-based scheduling, PIR-sensor analysis, and servo motor control.
- →Implemented control logic for the pump, wiper mechanism, LCD display, and relay module for system switching.
- →Integrated NodeMCU (IoT) with the Arduino using the Blynk mobile application to enable remote ON/OFF control.
- →Conducted testing and calibration of PIR sensor thresholds, servo angles, and timing to ensure reliable cleaning performance.
- →Assisted in hardware assembly, connection design, and overall prototype validation.
Academic Background
Education
April 2024 – Present
Master's in Big Data & Artificial Intelligence
SRH Hochschule Berlin
Berlin, Germany
2019 – 2023
Bachelor's in Electrical & Electronic Engineering
MVJCE
Bengaluru, India
Get In Touch
Let's work together
Available for full-time opportunities in AI, data engineering, and software development.
devappallyshivannavivek@gmail.com
Phone
+49 176 20801967
Location
Berlin, Germany
Connect with me
Languages
English (C1) • German (B1)
Interests
AI, Data Science, Emerging Technologies
Hobbies
Gym, Swimming, Extreme Sports