AI/ML Engineer

AI/ML Engineer - Local LLM & RAG Systems

PTFS is seeking an experienced AI/ML Engineer with strong expertise in deploying and managing locally hosted Large Language Models (LLMs) and building

Retrieval-Augmented Generation (RAG) pipelines. The ideal candidate has hands-on experience with frameworks such as Ollama, LangChain, LlamaIndex, or VLLM, and is highly skilled in Python-based orchestration, vector search, and scalable data storage systems such as Vector Databases or Apache Solr. This role will be responsible for designing, optimizing, and maintaining our on-premise or air-gapped GenAI infrastructure, integrating new models, and keeping our architecture modular and future-proof.

LLM Deployment & Orchestration

Deploy, run, and optimize locally hosted LLMs using frameworks such as Ollama, VLLM, GPT4All, or HuggingFace Transformers.

• Build and maintain model-serving pipelines with Python, including GPU optimization, quantization, batching, and model switching.
• Implement flexible architecture allowing rapid integration of new open-source or proprietary models.

RAG Pipeline Development

Architect end-to-end Retrieval-Augmented Generation (RAG) systems.

• Design and implement vector embedding, indexing, and retrieval layers, including chunking, metadata management, and routing logic.
• Integrate RAG flows using LangChain or LlamaIndex, ensuring low latency and high retrieval accuracy.

Data Storage and Retrieval

• Develop and maintain Vector Databases such as:
  • Pinecone
  • Weaviate
  • Chroma
  • Milvus
  • FAISS
• Or , architect a schema and search strategy for a Solr-based alternative using traditional indexing/search if vectors are not used.
• Manage ingestion pipelines, embedding generation, and update workflows for newly added data sources.

Application & API Development

Build backend services and APIs that interact with LLMs, embedding pipelines, and retrieval layers.

• Integrate agents, tools, and orchestration flows using:

• LangChain
• OpenAI function-calling equivalents in local models
• Custom Python toolchains

• Deploy services using Docker, Kubernetes, or local orchestrators when needed.

System Performance, Optimization & Monitoring

• Optimize model performance, including:
  • quantization (GGUF, GPTQ, AWQ)
  • tensor parallelization
  • caching strategies

• Monitor system resources for memory, GPU/CPU utilization, and throughput.
• Implement automated pipelines to update models, refresh embedding stores, and version datasets.

** Collaboration & Architecture**

• Work with cross-functional teams to align the LLM capabilities with business needs.
• Provide guidance on GenAI trends, limitations, and best practices.
• Contribute to documentation and provide internal training when needed.

Required Skills & Experience
Technical Skills

• 3-7+ years of experience in Machine Learning, MLOps, Backend Engineering, or AI Infrastructure.
• Expert-level proficiency in Python and relevant libraries (FastAPI, Pydantic, PyTorch, HuggingFace Ecosystem).
• Hands-on experience with LLM deployment via:
  • Ollama
  • VLLM
  • GPT4All
  • HuggingFace Transformers
  • LM Studio
• Strong experience with RAG frameworks:
  • LangChain
  • LlamaIndex
• Proficiency with vector databases (Pinecone, Chroma, Weaviate, FAISS, Milvus).
• Experience with Solr, Elasticsearch, or OpenSearch (schema design, analyzers, indexing).
• Experience developing embeddings pipelines, chunking strategies, and metadata retrieval.
• Familiarity with containerization and orchestration (Docker, Kubernetes optional).
• Strong experience with model inference optimization: quantization, batching, GPU acceleration.

ML/AI Knowledge

• Understanding of foundational LLM mechanics: transformers, tokenization, context windows, prompt engineering.
• Experience with model fine-tuning, LoRA adapters, or supervised fine-tuning (a plus).
• Knowledge of GenAI architectural patterns, agents, routing, tool use, and document indexing strategies.Preferred Qualifications Experience working in air-gapped or on-premise environments.
• Experience with CI/CD for ML systems.
• Familiarity with:
  • Nvidia GPU stack (CUDA, cuBLAS, TensorRT)
  • DevOps tools (Terraform, Ansible, Helm charts)
• Exposure to hybrid search systems combining vector + keyword retrieval (BM25 + embeddings).
• Experience integrating LLMs into enterprise systems.Education
• Bachelor's degree in Computer Science, Data Science, Engineering, Mathematics, or related field.
• Master's or higher preferred, but equivalent experience accepted.

Soft Skills

• Strong problem-solving ability and comfort working with ambiguous or evolving requirements.
• Excellent communication and ability to translate technical concepts for non-technical teams.
• Self-driven with a passion for exploring new GenAI technologies and keeping current with evolving LLM tools.

Summary

This role is ideal for someone who enjoys building practical, production-ready AI systems, particularly local LLMs, and wants to work at the cutting edge of the GenAI landscape-integrating models, designing robust retrieval systems, and ensuring future scalability.