AI & Autonomous Systems Engineer · Agentic AI · Computer Vision · Robotics & IoT
Braunschweig, Germany · patnisiddharth1311@gmail.com
I build production-grade LLM infrastructure, computer vision pipelines, and robotics integrations. My focus is on systems that act rather than only respond: agentic web automation with self-learning fallback cascades, LLM observability tooling, multi-robot orchestration, and visual workflow engines.
Currently pursuing an M.Sc. in Digital Technologies at TU Clausthal and Ostfalia University, Germany, alongside AI research and development work at Ciconia Systems GmbH.
Working languages: English (C1), German (A2–B1)
| Role | Organisation | Focus | Period |
|---|---|---|---|
| AI Software Engineer (R&D) | Ciconia Systems GmbH, CORE Research Group | Vergabepilot.AI, autonomous procurement document extraction | 2026 – Present |
| Open Source Developer | Independent | Spidey, Sentinel, Cereforge, Workflow Engine | 2025 – Present |
| Research Assistant | Academic Labs, IoT Prototyping | Robotics and embedded systems (ShopMate-R, AI Driver Safety) | 2024 – 2025 |
| Software Developer Intern | L&T Technology Services | React and Flask services, Docker CI/CD, ~30% latency reduction | Dec 2023 – Apr 2024 |
| Data Analyst Intern | Snapfix Ltd. | SQL automation and workflow auditing | Jun 2023 – Jul 2023 |
Private research project, CORE Research Group / Ciconia Systems
Autonomous agentic system for extracting structured data from public procurement portals across 30+ portal families. Uses a seven-stage fallback cascade (cached, deterministic, adaptive, LLM code generation, learned route, computer-use agent, manual) with replayable self-learned routes, Redis-backed circuit breakers, 32-field deep extraction, and a suite of 262 tests.
Python · FastAPI · Celery · Redis · PostgreSQL · MinIO · Playwright · browser-use · Docker
Self-hosted autonomous agent platform with a live React Flow reasoning graph, offline speech recognition via Vosk, local model execution through Ollama, and a two-stage SFT to DPO fine-tuning pipeline spanning 17 modules. Runs entirely offline.
Python · FastAPI · React · React Flow · Vosk · Ollama · Unsloth · TRL
LLM observability proxy featuring span-tree request waterfalls, judge-model evaluations, and a SHA-256 hash-chained audit ledger designed against EU AI Act record-keeping requirements.
FastAPI · Celery · Next.js
Competitive AI engineering platform with 24 curated tasks, automated evaluation via Gemini, and an XP and leaderboard progression system.
FastAPI · React · PostgreSQL
Distributed low-code automation platform in the spirit of n8n, built on Bull workers with sandboxed Node VM and containerised Python runtimes.
Node.js · React Flow · Redis
Multi-robot grocery assistant combining Pepper for dialogue, Temi for autonomous navigation, and M5Stack IoT shelf sensors for inventory awareness.
Python · Flask · Robotics
| Project | Description |
|---|---|
| AI Driver Safety | ADAS prototype: CNN traffic sign recognition, YOLOv4 pedestrian detection, dlib drowsiness monitoring, Arduino and OBD-II braking interface |
| Digital Inventory | Retail SaaS with QR-based lookup, audit logging, hardened access control, and investor pitch materials |
| LLM Chatbot | Self-hosted DialoGPT deployment optimised for Apple Silicon (MPS), containerised with Docker |
| SmartBot | Offline-first FAQ assistant built with Flask and React, requiring no external API calls |
| Multi-Disease Diagnostic | Medical image classification pipeline reaching approximately 85% accuracy and reducing per-case review from 30 minutes to 10 seconds |
Languages: Python, TypeScript, JavaScript, SQL
AI & Machine Learning: PyTorch, TensorFlow, Unsloth, TRL, SFT and DPO fine-tuning, Ollama, computer vision (CNN, YOLO, dlib)
Agentic Systems: browser-use (computer-use agents), Playwright automation, cascade orchestration, self-healing scrapers
Backend: FastAPI, Flask, Node.js, Celery, Bull, Redis, PostgreSQL, MinIO
Frontend: React, Next.js, React Flow, Tailwind CSS
Infrastructure: Docker, GitHub Actions, Prometheus, Vercel
Robotics & IoT: Pepper, Temi, Arduino, OBD-II, M5Stack
| Qualification | Institution | Period |
|---|---|---|
| M.Sc. Digital Technologies | TU Clausthal & Ostfalia University, Germany | Mar 2025 – Present |
| B.Tech Computer Engineering | CHARUSAT, India (CGPA 7.6/10) | 2021 – 2024 |
| Diploma in Computer Engineering | GTU, India (CGPA 8.92/10, High Distinction) | 2018 – 2021 |
- Email: patnisiddharth1311@gmail.com
- LinkedIn: siddharth-divyang-patni
- Portfolio: siddharth-portfolio-pi.vercel.app
Open to roles and collaboration in agentic AI, LLM infrastructure, and autonomous systems.
The portfolio site is a single-page Next.js 15 application (App Router) written in TypeScript with Tailwind CSS v4, deployed on Vercel.
npm install # install dependencies
npm run dev # start dev server at http://localhost:3000
npm run build # create a production buildStructure
app/— layout, page, and global stylescomponents/— page sections, UI primitives, and visual effectslib/data/— typed content for projects, experience, skills, and education