Gen AI Engineer

Gen AI Engineer

Contract : 12 Months

Experience : 5+ years

Location : Bangalore

Job Type- Hybrid/3 days office

Notice Period: 30 Days

Client is looking for an experienced Generative AI/LLM Engineer to design, build, and productionize LLM-powered applications. You will work across the stack—from prompt design and retrieval-augmented generation (RAG) to fine-tuning and scalable deployment—using platforms like OpenAI, Google Vertex AI, and Hugging Face, with frameworks such as Semantic Kernel and LangChain.

Key Responsibilities

• Design & Build LLM Solutions: Architect and implement GenAI features (chatbots, copilots, summarization, extraction, agents, content generation) using LangChain/Semantic Kernel.
• RAG Pipelines: Implement retrieval pipelines (chunking, embeddings, vector search) with FAISS, Pinecone, or Vertex Matching Engine; optimize grounding & hallucination control.
• Model Integration & Tuning: Integrate OpenAI/Vertex AI/Hugging Face models; evaluate fine-tuning vs. adapters (LoRA/QLoRA) on domain data.
• MLOps for GenAI: Package and deploy services (Docker, REST/gRPC), orchestrate workflows, monitor latency/cost/quality, manage keys/secrets/compliance.
• Data & Visualization: Analyze datasets, prompts, and outputs using Pandas, Matplotlib, and Seaborn to drive iterative improvements.
• Collaboration: Partner with Product, Data, and Platform teams; document design decisions, APIs, and runbooks.

Required Qualifications (4–7 years)

• Hands-on with LLM frameworks: LangChain and/or Semantic Kernel (agents, tools, memory, chains/planners).
• Model APIs & Platforms: Strong experience with OpenAI (GPT family & Assistants), Google Vertex AI (text-bison/gemini, Model Garden, Workbench), and Hugging Face (Transformers, Inference/PEFT).
• NLP & Python: Solid Python; experience with Transformers, tokenization, embeddings, vector stores; Matplotlib/Seaborn for EDA & evals.
• RAG & Prompting: Proven track record designing prompts, system messages, function/tool calling, RAG optimization, and cost/latency trade-offs.
• Deployment: Building scalable services (Docker, FastAPI/Flask, cloud functions); CI/CD familiarity.