Description
Summary Project Overview We are looking for an experienced AI/LLM Engineer and Technical Trainer to deliver an advanced training program for our engineering team focused on Enterprise LLMs, Retrieval-Augmented Generation (RAG), LoRA, and Efficient Fine-Tuning. The goal is to transform our technical team into experts capable of designing, deploying, and maintaining enterprise AI solutions using open-source LLMs. This engagement combines technical instruction, architecture guidance, hands-on labs, and mentoring. Training Scope Block 1 – Enterprise LLM & RAG Fundamentals Duration: 1–2 weeks Module 1: Embeddings & Vector Databases Topics include: How embeddings represent documents as vectors Semantic search concepts Vector similarity search Vector database architectures Storage options and tradeoffs Pinecone Weaviate Milvus Qdrant Indexing strategies Chunking techniques Metadata filtering Module 2: Retrieval & Generation Evaluation Teach best practices for evaluating RAG systems, including: Retrieval metrics NDCG MRR Recall Precision Generation quality metrics BLEU ROUGE Hallucination detection RAG benchmarking Evaluation pipelines Ground-truth datasets Module 3: Build an Internal Enterprise Chatbot (POC) Guide the team through building a production-style proof of concept: Enterprise documentation chatbot Private knowledge base Source citation and traceability Document ingestion pipeline RAG architecture Deployment within existing infrastructure Module 4: RAG Integration Platform Explain how to orchestrate RAG pipelines using a centralized platform, including: Model orchestration Prompt pipelines Retrieval workflows API integration Enterprise architecture Observability Block 2 – LoRA & Efficient Fine-Tuning Duration: 2–3 weeks Module 1: LoRA & QLoRA Fundamentals Topics include: Parameter-efficient fine-tuning LoRA architecture QLoRA Adapter-based training Low-rank matrices GPU memory optimization VRAM reduction Cost comparison: LoRA vs Full Fine-Tuning Performance tradeoffs Module 2: Dataset Preparation Teach best practices for: Dataset collection Cleaning Labeling Validation Balancing Support conversations FAQs Domain-specific documentation Synthetic data generation Module 3: Domain-Specific Adapter Training Hands-on training covering: Training LoRA adapters Financial domain Telecommunications Legal Healthcare Model evaluation Comparison against base models Module 4: Reusable Adapter Catalog Teach how to build reusable adapter libraries: Modular adapters Domain-specific adapters Adapter versioning Combining LoRA with RAG Multi-domain architecture Production deployment Desired Deliverables The selected expert will: Deliver live remote training sessions Prepare presentation materials Create hands-on labs Provide code examples Assist with architecture discussions Review participant exercises Answer technical questions Help build a production-quality Proof of Concept (POC) Required Skills We are looking for someone with strong experience in: LLMs Llama Mistral Qwen Gemma DeepSeek Open-source language models RAG LangChain LlamaIndex Haystack Hybrid Search Vector databases Embeddings Vector Databases Experience with one or more: Pinecone Weaviate Milvus Qdrant Fine-Tuning LoRA QLoRA PEFT Hugging Face Transformers TRL Unsloth (preferred) Axolotl (nice to have) ML Frameworks PyTorch Hugging Face Accelerate BitsAndBytes Deployment FastAPI vLLM TGI Docker Kubernetes (preferred) REST/gRPC APIs Evaluation Experience with: RAG evaluation BLEU ROUGE NDCG MRR Hallucination detection Benchmarking Nice to Have Enterprise AI architecture experience Production LLM deployment Multi-tenant AI platforms GPU optimization Model serving at scale Experience teaching engineers or delivering corporate training English fluency (Spanish is a strong plus) Engagement Remote Part-time / Contract Approximately 3–5 weeks Live sessions plus office hours Flexible scheduling To Apply Please include: A brief introduction about your experience. Examples of enterprise RAG or LLM projects you've built. Experience with LoRA/QLoRA fine-tuning. Experience teaching or mentoring technical teams. Links to GitHub, technical blog posts, publications, or conference talks (if available). Your proposed hourly rate and availability.