LLM infrastructure
PocketLLM ecosystem
Cloud-connected and local-first assistants built around a shared AI backend.
- Python
- FastAPI
- Flutter
- Ollama

AI engineer · Trier, Germany
Building useful AI, from the model
to the product people use.
Python · FastAPI · LLM & agentic systems
M.Sc. Natural Language Processing, University of Trier
A little about me
Hello, I’m Prashant.
AI/ML engineer and M.Sc. Natural Language Processing student at the University of Trier. I build Python/FastAPI backends, retrieval systems and agentic workflows, with experience taking AI products from data ingestion and orchestration to production integration.
I work across retrieval, model evaluation, APIs and the interfaces people use to inspect results. My core stack is Python, FastAPI, PyTorch and Hugging Face, with LangGraph for stateful workflows and PostgreSQL for durable data. I value reproducible experiments, clear failure states and software that can be maintained after the demo.
I’m currently looking for student-compatible opportunities in AI/ML, Python backend development and NLP.

01 / Where I’ve contributed
AI systems in production.
A practical eye for the details.
Built Python/FastAPI AI backends, production RAG pipelines, agentic proposal-generation workflows and voice/IVR integrations. Contributed website ingestion and vector retrieval for grounded proposal generation.
Worked on retrieval, prompting, context management and provider integration for multi-step AI workflows.
September 2023–January 2024
Developed conversational AI prototypes with Python, TensorFlow, Rasa and Dialogflow; evaluated intent and dialogue workflows.
Reported more than 150 product issues and edge cases.
02 / Always learning
From applied AI
to the language behind it.
Postgraduate · In progress
University of Trier
April 2026 – expected September 2028
Trier, Germany
Undergraduate · Completed
Uka Tarsadia University
2021–2025
Bardoli, India03 / Ideas made tangible
LLM infrastructure
Cloud-connected and local-first assistants built around a shared AI backend.
NLP research
Reproducing few-shot classification and investigating a paper-to-code discrepancy.
Language model engineering
A compact decoder-only language-model stack for constrained consumer GPUs.
Applied AI product
An AI-assisted job application workflow with reviewable changes and application tracking.
Loading projects from GitHub...
My technical arsenal for building intelligent solutions.




04 / Let’s make something useful
Have a role in mind, a technical question or a product you want to build? Choose the detail that suits your idea.
A quick hello
Have a question or want to work together? I’d love to hear from you. Send me a message and I’ll respond as soon as possible.
A little more detail
Share the goals, scope and timeline. We’ll start with what matters most.
Let's start with some basic information
Powered by GPT OSS 120B
I can answer questions about his projects, skills, experience, education, and public profile links.