AI engineer focused on LLM apps, agentic workflows, and automation systems, with strong software engineering and safety-critical systems experience.
Aktualisiert am 27.04.2026
Profil
Freiberufler / Selbstständiger
Remote-Arbeit
Verfügbar ab: 01.05.2026
Verfügbar zu: 100%
davon vor Ort: 100%
Python
Agentic AI
Automation
Machine Learning / AI (LLMs & RAG)
LangChain
LangGraph
RAG (Retrieval-Augmented Generation)
n8n
FastAPI
Docker
AWS (EC2
RDS)
PostgreSQL
REST APIs
Prompt Engineering
Context Engineering
Multi-Agent Systems
LLM Orchestration
Test Automation
CI/CD
GitHub
Vector databases
German
Intermediate Knowledge B1
English
Professional Proficiency C1
Telugu
Native C2
Hindi
Intermediate Knowledge B1

Einsatzorte

Einsatzorte

Regensburg (+500km) Munich (+500km) Berlin (+500km) Frankfurt am Main (+500km) Ulm (Donau) (+500km)
Deutschland
möglich

Projekte

Projekte

2 months
2026-03 - now

OpenToWork: Autonomous Multi-Agent Job Application System

  • Architected and deployed a 6-agent autonomous job application system orchestrated via n8n, with each agent implemented as an independent FastAPI microservice and backed by PostgreSQL for persistent state and workflow chaining.
  • Developed LLM-driven agent workflows for CV?job matching, skill gap analysis, automated interview preparation and ATS-optimized CV and cover letter generation using dynamic keyword injection.
  • Deployed a production-grade agentic system on AWS using Docker Compose (n8n + FastAPI), EC2 and RDS PostgreSQL, exposing a 15-endpoint API layer and processing 1000+ real job postings end-to-end without human intervention.
3 months
2026-02 - now

Multilingual ASR & Subtitle Generation System

  • Fine-tuned Whisper ASR model for Telugu?English code-switching using a custom dataset, reducing Word Error Rate from 75% to 15.1% and publishing the model to HuggingFace Hub.
  • Built an end-to-end MLOps and deployment pipeline using MLflow for experiment tracking and model versioning and deployed a production inference service using FastAPI with Prometheus and Grafana monitoring, containerized via Docker Compose.
  • Improved inference performance using CTranslate2 int8 quantization, achieving 4?5× faster CPU transcription and deploying the system as CLI, web app and API with subtitle export.
autoTinglishSub
8 months
2025-09 - now

RAG-Based Personal AI Assistant & Voice Agent

  • Built a RAG pipeline using Mistral mistral-embed (1024-dim) embeddings and Supabase PGVector for semantic document retrieval, with a semantic response cache (cosine similarity ? 0.92) that bypasses LLM calls on repeated queries.
  • Implemented LLM-based intent classification and communication style detection using LangGraph, dynamically adapting CTA wording and response style per user.
  • Built a real-time voice-enabled conversational AI system using a voice-cloned TTS model and Web Speech API VAD/STT, enabling low-latency speech-to-response interaction with idle nudge handling and audio queue management.
talkToVasu
11 months
2025-06 - now

Multi-Agent Financial Decision System

  • Architected and deployed a production multi-agent AI system using a custom GAME (Goal?Action - Memory?Environment) framework, implementing specialized financial agents with isolated tool registries, memory stores and execution environments powered by LLM APIs.
  • Built an agent orchestration layer enabling agent-to-agent communication, intent classification and dynamic task routing between autonomous agents.
  • Designed a persistent agent memory system using BM25 retrieval over historical decisions and outcomes, enabling retrieval-based reasoning and experiential learning without vector embeddings.
  • Deployed the system on AWS (EC2, RDS) with FastAPI microservices, CI/CD via GitHub Actions, automated test suite (250+ tests) and bidirectional MCP integration for standardized tool access across agents.
FinsenseAI
1 year 1 month
2025-04 - now

Multi-Agent Nutrition Analysis System (RAG + LLM)

  • Developed an LLM-powered analysis system using FastAPI, LangChain, and Mistral 7B, integrating Open-FoodFacts (3M+ products) to generate structured nutrition and ingredient analysis.
  • Designed a multi-agent RAG architecture with specialized agents for ingredient analysis, nutrition scoring, regulatory classification, and response synthesis.
  • Implemented advanced RAG techniques including Chain-of-Thought reasoning, Corrective RAG (CRAG) for low-confidence retrieval, and Multi-Hop retrieval over a regulatory knowledge base stored in ChromaDB.
beyondLabel
1 month
2020-12 - 2020-12

Self Driving Toy Car

  • Designed and developed complete hardware for a self-driving car using Raspberry Pi, Raspberry Pi Camera,  Arduino UNO and Li-Po battery.
  • Implemented lane detection using OpenCV image analysis techniques to identify lanes.
  • Trained an image classifier to detect objects, stop signs, and traffic lights.
6 months
2017-11 - 2018-04

Preparation of the bachelor's thesis

  • Developed a real-time driver drowsiness detection system using computer vision and image processing techniques.
  • Implemented face detection and facial landmark extraction to monitor driver fatigue indicators.
  • Built an image processing pipeline for continuous driver state monitoring and fatigue detection.
Vignan?s Institute of Information Technology, Visakhapatnam (India)

Aus- und Weiterbildung

Aus- und Weiterbildung

2 years 10 months
2018-10 - 2021-07

Automotive Software Engineering

Master of Science, Technische Universität Chemnitz, Chemnitz (Germany)
Master of Science
Technische Universität Chemnitz, Chemnitz (Germany)
3 years 9 months
2014-08 - 2018-04

Computer Science

Bachelor of Technology, Vignan?s Institute of Information Technology, Visakhapatnam (India)
Bachelor of Technology
Vignan?s Institute of Information Technology, Visakhapatnam (India)

Kompetenzen

Kompetenzen

Top-Skills

Python Agentic AI Automation Machine Learning / AI (LLMs & RAG) LangChain LangGraph RAG (Retrieval-Augmented Generation) n8n FastAPI Docker AWS (EC2 RDS) PostgreSQL REST APIs Prompt Engineering Context Engineering Multi-Agent Systems LLM Orchestration Test Automation CI/CD GitHub Vector databases

Schwerpunkte

Ability to Work independently
Problem solving
Team player
Motivated
Responsible

Produkte / Standards / Erfahrungen / Methoden

PROFILE
AI Engineer with 4+ years of experience specializing in building production-grade AI systems, including multiagent architectures, RAG pipelines and intelligent workflow automation. Experienced in deploying LLMpowered systems using LangChain, Whisper, MCP and n8n on AWS with real-world data and agentic decisionmaking.

SKILLS
  • AI/ Machine
    • Learning Machine Learning
    • Deep Learning
    • NLP
    • Speech Recognition
    • Computer Vision
    • PyTorch
    • TensorFlow
    • Scikit-learn
    • Model Fine-Tuning
    • Whisper 
    • Mistral (Embeddings)
  • Agentic AI/ Generative AI 
    • LLMs 
    • Multi-Agent Systems
    • RAG
    • Agent Orchestration
    • Tool-Using Agents
    • Agent Memory Systems
    • LangChain 
    • LangGraph 
    • Context Engineering 
    • Prompt Engineering 
    • Tool-Using Agents 
    • MCP
  • Backend & Cloud
    • FastAPI 
    • REST APIs
    • Microservices
    • System Design AWS (EC2, RDS)
    • Docker 
    • Docker Compose
    • CI/CD (GitHub Actions) 
    • MLflow 
    • DVC
    • Prometheus
    • Grafana
    • n8n 
    • Supabase
  • Tools
    • Git
    • ?Linux

WORK EXPERIENCE
09/2021 ? 02/2026
Software Engineer - AI & Test Automation
EDAG Group, Regensburg (Germany)
  • Designed and implemented an AI-powered test generation system that converts structured requirements into automated test cases (CMOCKA/Python/C++), reducing manual effort by up to 80%.
  • Developed a RAG-based pipeline using FAISS/PGVector to reuse historical test data, improving test coverage with automated edge-case and boundary-condition generation.
  • Implemented a multi-agent AI system with context engineering and MCP tool integration, enabling dynamic test generation, validation, and execution within a FastAPI-based architecture.
  • Built and automated Model-in-the-Loop (MiL) and Software-in-the-Loop (SiL) test environments using cmocka and Python, improving test efficiency and enabling scalable validation workflows.
  • Contributed to CI/CD-driven test automation and ensured traceability across requirements, test cases and validation artifacts in compliance with ASPICE and ISO 26262

09/2020 ? 07/2021
Master Student
IAV GmbH, Munich (Germany)
Lithium-Ion Battery State of Charge Modelling based on Neural Networks
  • Developed a data-driven approach for modeling a Lithium-ion battery by using Neural Networks.
  • Built a data pipeline (Data collection, Preprocessing, Training, Validation) which represents preprocessing to actual training of NN?s and validating the results.
  • Validated models with reference measurements and compared with other model architectures and as well as to conventional battery models.

04/2020 ? 09/2020
IAV GmbH, Munich (Germany)
Research Internship
Battery Modelling with different Machine Learning processes
  • Developed a simulation environment which provides a battery model from the data.
  • Implemented data pipeline which makes building models more efficient and simplified.
  • Trained models and created test data sets to validate the effectiveness of the algorithms and to ensure that the models are accurate and efficient.

Programmiersprachen

Python
C
C++
Embedded C
SQL
HTML

Datenbanken

PostgreSQL
SQL
Vector Databases
FAISS, ChromaDB, PGVector
Data Pipelines
Data Preprocessing
ETL

Einsatzorte

Einsatzorte

Regensburg (+500km) Munich (+500km) Berlin (+500km) Frankfurt am Main (+500km) Ulm (Donau) (+500km)
Deutschland
möglich

Projekte

Projekte

2 months
2026-03 - now

OpenToWork: Autonomous Multi-Agent Job Application System

  • Architected and deployed a 6-agent autonomous job application system orchestrated via n8n, with each agent implemented as an independent FastAPI microservice and backed by PostgreSQL for persistent state and workflow chaining.
  • Developed LLM-driven agent workflows for CV?job matching, skill gap analysis, automated interview preparation and ATS-optimized CV and cover letter generation using dynamic keyword injection.
  • Deployed a production-grade agentic system on AWS using Docker Compose (n8n + FastAPI), EC2 and RDS PostgreSQL, exposing a 15-endpoint API layer and processing 1000+ real job postings end-to-end without human intervention.
3 months
2026-02 - now

Multilingual ASR & Subtitle Generation System

  • Fine-tuned Whisper ASR model for Telugu?English code-switching using a custom dataset, reducing Word Error Rate from 75% to 15.1% and publishing the model to HuggingFace Hub.
  • Built an end-to-end MLOps and deployment pipeline using MLflow for experiment tracking and model versioning and deployed a production inference service using FastAPI with Prometheus and Grafana monitoring, containerized via Docker Compose.
  • Improved inference performance using CTranslate2 int8 quantization, achieving 4?5× faster CPU transcription and deploying the system as CLI, web app and API with subtitle export.
autoTinglishSub
8 months
2025-09 - now

RAG-Based Personal AI Assistant & Voice Agent

  • Built a RAG pipeline using Mistral mistral-embed (1024-dim) embeddings and Supabase PGVector for semantic document retrieval, with a semantic response cache (cosine similarity ? 0.92) that bypasses LLM calls on repeated queries.
  • Implemented LLM-based intent classification and communication style detection using LangGraph, dynamically adapting CTA wording and response style per user.
  • Built a real-time voice-enabled conversational AI system using a voice-cloned TTS model and Web Speech API VAD/STT, enabling low-latency speech-to-response interaction with idle nudge handling and audio queue management.
talkToVasu
11 months
2025-06 - now

Multi-Agent Financial Decision System

  • Architected and deployed a production multi-agent AI system using a custom GAME (Goal?Action - Memory?Environment) framework, implementing specialized financial agents with isolated tool registries, memory stores and execution environments powered by LLM APIs.
  • Built an agent orchestration layer enabling agent-to-agent communication, intent classification and dynamic task routing between autonomous agents.
  • Designed a persistent agent memory system using BM25 retrieval over historical decisions and outcomes, enabling retrieval-based reasoning and experiential learning without vector embeddings.
  • Deployed the system on AWS (EC2, RDS) with FastAPI microservices, CI/CD via GitHub Actions, automated test suite (250+ tests) and bidirectional MCP integration for standardized tool access across agents.
FinsenseAI
1 year 1 month
2025-04 - now

Multi-Agent Nutrition Analysis System (RAG + LLM)

  • Developed an LLM-powered analysis system using FastAPI, LangChain, and Mistral 7B, integrating Open-FoodFacts (3M+ products) to generate structured nutrition and ingredient analysis.
  • Designed a multi-agent RAG architecture with specialized agents for ingredient analysis, nutrition scoring, regulatory classification, and response synthesis.
  • Implemented advanced RAG techniques including Chain-of-Thought reasoning, Corrective RAG (CRAG) for low-confidence retrieval, and Multi-Hop retrieval over a regulatory knowledge base stored in ChromaDB.
beyondLabel
1 month
2020-12 - 2020-12

Self Driving Toy Car

  • Designed and developed complete hardware for a self-driving car using Raspberry Pi, Raspberry Pi Camera,  Arduino UNO and Li-Po battery.
  • Implemented lane detection using OpenCV image analysis techniques to identify lanes.
  • Trained an image classifier to detect objects, stop signs, and traffic lights.
6 months
2017-11 - 2018-04

Preparation of the bachelor's thesis

  • Developed a real-time driver drowsiness detection system using computer vision and image processing techniques.
  • Implemented face detection and facial landmark extraction to monitor driver fatigue indicators.
  • Built an image processing pipeline for continuous driver state monitoring and fatigue detection.
Vignan?s Institute of Information Technology, Visakhapatnam (India)

Aus- und Weiterbildung

Aus- und Weiterbildung

2 years 10 months
2018-10 - 2021-07

Automotive Software Engineering

Master of Science, Technische Universität Chemnitz, Chemnitz (Germany)
Master of Science
Technische Universität Chemnitz, Chemnitz (Germany)
3 years 9 months
2014-08 - 2018-04

Computer Science

Bachelor of Technology, Vignan?s Institute of Information Technology, Visakhapatnam (India)
Bachelor of Technology
Vignan?s Institute of Information Technology, Visakhapatnam (India)

Kompetenzen

Kompetenzen

Top-Skills

Python Agentic AI Automation Machine Learning / AI (LLMs & RAG) LangChain LangGraph RAG (Retrieval-Augmented Generation) n8n FastAPI Docker AWS (EC2 RDS) PostgreSQL REST APIs Prompt Engineering Context Engineering Multi-Agent Systems LLM Orchestration Test Automation CI/CD GitHub Vector databases

Schwerpunkte

Ability to Work independently
Problem solving
Team player
Motivated
Responsible

Produkte / Standards / Erfahrungen / Methoden

PROFILE
AI Engineer with 4+ years of experience specializing in building production-grade AI systems, including multiagent architectures, RAG pipelines and intelligent workflow automation. Experienced in deploying LLMpowered systems using LangChain, Whisper, MCP and n8n on AWS with real-world data and agentic decisionmaking.

SKILLS
  • AI/ Machine
    • Learning Machine Learning
    • Deep Learning
    • NLP
    • Speech Recognition
    • Computer Vision
    • PyTorch
    • TensorFlow
    • Scikit-learn
    • Model Fine-Tuning
    • Whisper 
    • Mistral (Embeddings)
  • Agentic AI/ Generative AI 
    • LLMs 
    • Multi-Agent Systems
    • RAG
    • Agent Orchestration
    • Tool-Using Agents
    • Agent Memory Systems
    • LangChain 
    • LangGraph 
    • Context Engineering 
    • Prompt Engineering 
    • Tool-Using Agents 
    • MCP
  • Backend & Cloud
    • FastAPI 
    • REST APIs
    • Microservices
    • System Design AWS (EC2, RDS)
    • Docker 
    • Docker Compose
    • CI/CD (GitHub Actions) 
    • MLflow 
    • DVC
    • Prometheus
    • Grafana
    • n8n 
    • Supabase
  • Tools
    • Git
    • ?Linux

WORK EXPERIENCE
09/2021 ? 02/2026
Software Engineer - AI & Test Automation
EDAG Group, Regensburg (Germany)
  • Designed and implemented an AI-powered test generation system that converts structured requirements into automated test cases (CMOCKA/Python/C++), reducing manual effort by up to 80%.
  • Developed a RAG-based pipeline using FAISS/PGVector to reuse historical test data, improving test coverage with automated edge-case and boundary-condition generation.
  • Implemented a multi-agent AI system with context engineering and MCP tool integration, enabling dynamic test generation, validation, and execution within a FastAPI-based architecture.
  • Built and automated Model-in-the-Loop (MiL) and Software-in-the-Loop (SiL) test environments using cmocka and Python, improving test efficiency and enabling scalable validation workflows.
  • Contributed to CI/CD-driven test automation and ensured traceability across requirements, test cases and validation artifacts in compliance with ASPICE and ISO 26262

09/2020 ? 07/2021
Master Student
IAV GmbH, Munich (Germany)
Lithium-Ion Battery State of Charge Modelling based on Neural Networks
  • Developed a data-driven approach for modeling a Lithium-ion battery by using Neural Networks.
  • Built a data pipeline (Data collection, Preprocessing, Training, Validation) which represents preprocessing to actual training of NN?s and validating the results.
  • Validated models with reference measurements and compared with other model architectures and as well as to conventional battery models.

04/2020 ? 09/2020
IAV GmbH, Munich (Germany)
Research Internship
Battery Modelling with different Machine Learning processes
  • Developed a simulation environment which provides a battery model from the data.
  • Implemented data pipeline which makes building models more efficient and simplified.
  • Trained models and created test data sets to validate the effectiveness of the algorithms and to ensure that the models are accurate and efficient.

Programmiersprachen

Python
C
C++
Embedded C
SQL
HTML

Datenbanken

PostgreSQL
SQL
Vector Databases
FAISS, ChromaDB, PGVector
Data Pipelines
Data Preprocessing
ETL

Vertrauen Sie auf Randstad

Im Bereich Freelancing
Im Bereich Arbeitnehmerüberlassung / Personalvermittlung

Fragen?

Rufen Sie uns an +49 89 500316-300 oder schreiben Sie uns:

Das Freelancer-Portal

Direktester geht's nicht! Ganz einfach Freelancer finden und direkt Kontakt aufnehmen.