AI Engineer focused on building things that actually work in production.
I work on LLMs - fine-tuning them, aligning them with RLHF, wrapping them in agentic pipelines, and deploying them where latency and privacy constraints are real. Currently doing my Master's thesis on agentic AI at Hochschule Hof (Germany) while working as a GenAI engineer at Infinite Mind GmbH.
- At Infinite Mind - On-premise RLHF pipelines (privacy-first, no external APIs) and production RAG systems with hybrid retrieval, reranking, and real-time monitoring for enterprise clients
- Exploring - Multi-agent orchestration(Architect), LLM evaluation frameworks, reasoning in constrained environments
AskToYaana - Fine-tuned LLaMA 3.3 for domain-specific product recommendations. LSTM intent classifier, Django backend, GROQ inference.
LlamaIndex QA Engine - Hybrid doc Q&A combining BM25 + FAISS dense retrieval, LangChain agent orchestration, GROQ for real-time answers.
MLOps Pipeline - End-to-end MLOps with MLflow, Docker, GitHub Actions, DagsHub. Cut deployment time by 40%.
BERT + LSTM Sentiment Engine - Hybrid NLP pipeline, served as a FastAPI microservice for customer feedback classification.
LLMs / GenAI → HuggingFace, LangChain, LlamaIndex, Ollama, CrewAI, GROQ
Fine-tuning → PyTorch, TRL, RLHF (PPO + reward modeling), LoRA / QLoRA
RAG → FAISS, BM25, vector DBs, rerankers, real-time eval
MLOps → MLflow, W&B, Docker, Kubernetes, GitHub Actions, FastAPI
Cloud → AWS (SageMaker, Bedrock, Lambda, EC2), Azure OpenAI, Azure ML
1 year as an AI Engineer in India (Woxxin Technolies Ltd) - owned AI products end-to-end, client-facing, hands-on Azure.
B.E. IT from SCET Surat → M.Sc. AI & Robotics at Hochschule Hof, Germany.
AWS Certified AI Practitioner. 2nd place at ML Hackathon 2.0.
Full-time AI/ML Engineer roles in Germany or remote.
Dhananipreet101@gmail.com · LinkedIn · HuggingFace