An AI engineer
& full-stack builder.
AI Full Stack Engineer with 5+ years delivering production-grade LLM- and RAG-powered systems at Carl Zeiss, Barclays, and Hexagon. M.Sc. in Natural Language Processing — University of Trier.
Currently available for full-time roles · Germany 🇩🇪
Open to new opportunities in AI/ML & Full-Stack Engineering
AI Engineer with Full-Stack Experience
AI Full Stack Engineer with 5+ years delivering production-grade applications end-to-end — Python/FastAPI backends, React/TypeScript frontends, and REST API architectures in enterprise-regulated environments. Built an LLM-powered Microscopy Copilot at Carl Zeiss on Azure; previously shipped high-reliability transactional systems at Barclays (99.9% uptime) and industrial platforms at Hexagon (500+ users). Combines deep AI/ML expertise (RAG, multi-agent systems, LLM orchestration) with strong DevSecOps practices across Docker, CI/CD, and Azure. Proven track record leading code reviews, collaborating with vendor teams, and driving AI-assisted development adoption. M.Sc. in NLP (University of Trier, thesis completed).
My Tech Stack
Technical arsenal
Core Skills & Tools
A comprehensive toolkit honed over years of building resilient AI systems, full-stack platforms, and data pipelines.
Backend Development
Frontend Development
AI / Machine Learning
Frameworks & Tools
Cloud & DevSecOps
Databases
Spoken Languages
Career journey
Experience
Carl Zeiss Microscopy – Generative AI Engineer (Werkstudent)
- Took an LLM-driven Microscopy Copilot from prototype to production using Azure OpenAI and LangGraph, including evaluation frameworks, deployment pipelines, and reliability monitoring—reduced manual effort by 25–35%.
- ⭐ Named InventorZeiss AI Invention Disclosure (2025) for an agentic copilot system with human-review escalation points.
- Designed RAG pipelines with structured outputs and quality evaluation, integrating enterprise documentation and code repositories; improved retrieval accuracy by 40%.
- Built multi-step agentic workflows with tool-calling via MCP-based interoperability, connecting AI services to internal APIs, data sources, and operational processes across departments.
Barclays – Senior Software Developer
- Led system design of enterprise web applications for investment banking, building scalable RESTful APIs with async processing, automated testing, and delivery pipelines—achieving 99.9% uptime.
- Developed an internal AI-enabled workflow assistant automating employee support; defined escalation logic for policy dependent decisions, reducing routine tickets by 30%.
Hexagon CCI – Software Developer
- Delivered production industrial applications serving 5,000+ enterprise users; owned API design, data modelling, observability, and performance optimization (20% latency reduction).
Academic roots
Education
M.Sc. Natural Language Processing – University of Trier
- Thesis: Comparative Study of NLP Models for Analysis of AI/ML Skill Demand in the German Job Market
- Collected and annotated 1,000+ German job postings; benchmarked BERT-based, rule-based and LLM-based skill extraction models using Precision, Recall and F1-score for longitudinal trend analysis.
- Hands-on experience with Python, TensorFlow, PyTorch, HuggingFace, and deep learning frameworks for text classification, sentiment analysis, and conversational AI.
B.E. Computer Science – Chandigarh University
- Built strong foundations in algorithms, data structures, databases, and software engineering using Java, C++, and Python.
- Gained practical experience with AWS and cloud automation tools for deploying and managing scalable applications.
Certifications
- ✦LangChain Academy – Agentic AI (2026)
- ✦HuggingFace AI Agents Course (2026)
- ✦Microsoft AI Tour – GenAI & Azure (2025)
AI-Assisted Dev Tools
Selected work
Selected Work
A showcase of projects that highlight my commitment to building intelligent, production-grade systems.
Lensr – Intent-Aware AI Search Engine
Built a full-stack agentic search platform with 10 intents, 20 tailored result cards, and live SSE streaming using LangGraph, FastAPI, AWS Bedrock, React 19, and Supabase.
DailyAI – AI News Intelligence Platform
Shipped a production AI news app with source-backed summaries and personalization using FastAPI, NiceGUI, and LangGraph agentic pipelines; approx. 300 users, 40% retention.
NLP Skill Extraction – M.Sc. Thesis
Research-grade NLP pipeline comparing BERT fine-tuning, rule-based systems and GPT-4 for extracting AI/ML skills from 1,000+ real German job postings. Achieved state-of-the-art F1 on a custom annotated dataset, enabling longitudinal market trend analysis across 3 years of job-market data.
Agent-Based Research Assistant
Full-stack multi-agent system with FastAPI backend and React frontend. Features RAG-powered retrieval and real-time source citations for research queries. Agents autonomously plan, delegate, and execute complex multi-step reasoning tasks.
Multi-Model AI Chatbot
Sleek React chatbot that lets users switch between GPT-4, Claude and Gemini in real time. Features streaming responses, conversation memory and a polished UI. Handles 1,000+ messages per session with zero latency degradation. Deployed and live on Vercel.
LLM Document QA System
Document QA system with FAISS vector indexing, REST API layer, and generative LMs for grounded answers with source attribution. Upload any PDF or text file and get accurate, cited answers in seconds.
AI German Language Tutor
Conversational AI tutor that teaches German through adaptive dialogue, grammar correction and contextual vocabulary drills. Supports 10+ native languages as the instruction language. Used by learners across 5 countries. Deployed live on Streamlit Cloud.
AI Cover Letter Generator
Drop in a job description and your resume — get a tailored, ATS-optimised cover letter in under 10 seconds. Uses prompt-chaining to match tone, keywords and role requirements.
Automated Resume Screening Pipeline
End-to-end NLP pipeline that parses resumes, scores them against job descriptions using semantic similarity and ranks candidates automatically. Designed to cut recruiter screening time from hours to minutes.
Article: How I Built DailyAI With AI Coding Tools
Published in Towards AI (May 2026), detailing the architecture, prompt engineering, and agentic workflows used to build the DailyAI news platform.
Endorsements
Trusted by colleagues & collaborators
Abhishek Diwan
"Working with Shivang has been an incredible experience. His expertise in NLP and full-stack development was evident in our projects at Barclays, where he significantly improved the accuracy and efficiency of our NLP models. Shivang's proactive approach and strong collaboration skills made him an invaluable team member, consistently delivering high-quality work."
Durga Prasad Palle
"Shivang's contribution to the Hexagon OnCall Dispatch project was outstanding. His ability to architect and manage responsive web applications, coupled with his skill in optimizing performance, greatly enhanced our product. His dedication to understanding client needs and translating them into effective solutions was impressive, showcasing his strong technical and problem-solving skills."
Pooja Parakh
"Shivang is a highly skilled and dedicated IT professional. His work on deploying NLP models into production at Barclays demonstrated his deep technical knowledge and practical experience. Shivang's strong work ethic, attention to detail, and ability to work efficiently in a team make him a remarkable asset in any project or organization"
Prarthana Mukherjee
"Working with Shivang at Barclays was a great experience. His expertise in deploying NLP models into production was instrumental in improving our system's performance. Shivang's ability to handle complex technical challenges and deliver high-quality solutions under tight deadlines was impressive. His collaborative spirit and proactive approach to problem-solving made him an invaluable team member."
Abhishek Diwan
"Working with Shivang has been an incredible experience. His expertise in NLP and full-stack development was evident in our projects at Barclays, where he significantly improved the accuracy and efficiency of our NLP models. Shivang's proactive approach and strong collaboration skills made him an invaluable team member, consistently delivering high-quality work."
Durga Prasad Palle
"Shivang's contribution to the Hexagon OnCall Dispatch project was outstanding. His ability to architect and manage responsive web applications, coupled with his skill in optimizing performance, greatly enhanced our product. His dedication to understanding client needs and translating them into effective solutions was impressive, showcasing his strong technical and problem-solving skills."
Pooja Parakh
"Shivang is a highly skilled and dedicated IT professional. His work on deploying NLP models into production at Barclays demonstrated his deep technical knowledge and practical experience. Shivang's strong work ethic, attention to detail, and ability to work efficiently in a team make him a remarkable asset in any project or organization"
Prarthana Mukherjee
"Working with Shivang at Barclays was a great experience. His expertise in deploying NLP models into production was instrumental in improving our system's performance. Shivang's ability to handle complex technical challenges and deliver high-quality solutions under tight deadlines was impressive. His collaborative spirit and proactive approach to problem-solving made him an invaluable team member."
Get in touch.
Let's connect! I'm always open to discussing new opportunities, AI projects, or collaboration ideas.