AI Engineer

Job Title: AI Engineer
Location: -Viman Nagar, Pune
Experience- 4 plus Years
Timings: -11 am to 8 pm
Mode:- Hybrid Model

Be part of the AcquireX team that unleashes the power of leading-edge technologies to help improve e-commerce processes in the hardware world.

Job Description:-

Key Responsibilities

• Work on the full Gen-AI Stack: Optimize our full-stack, leveraging spec-driven development, Next.js + TypeScript, Python, and managed services like Azure OpenAI.
• Develop AI Systems: Design, develop, and deploy sophisticated AI Systems like Retrieval-Augmented Generation (RAG) to power our next-generation search and discovery experience.
• Develop & Fine-Tune LLMs: Develop advanced generative models for nuanced tasks like automated content creation, summarization, and metadata enrichment.
• Implement LLMOps: Embed robust observability using tools like OpenTelemetry and Prometheus. This includes tracking standard metrics (latency, cost, accuracy) and specialized monitoring for hallucination, toxicity, and data drift.

Must-Have Skills

• Production LLM Experience: Experience in full-stack development with demonstrable success in productionizing LLM applications using modern frameworks like DSPY, LangChain, LlamaIndex, or Hugging Face Transformers.
• RAG Expertise: Deep, practical knowledge of RAG architecture, including advanced prompt engineering, chunking strategies, and proficiency with vector databases (e.g., Pinecone, Weaviate, Milvus).
• Cloud Proficiency: Expertise with managed LLM services (Azure OpenAI Service or AWS Bedrock). Strong foundational cloud skills in either Azure or AWS for compute orchestration (AKS/EKS), serverless functions, and storage.
• MLOps Acumen: Solid experience with Docker, CI/CD pipelines (e.g., GitHub Actions, Argo), and model registries.

Nice-to-Have Skills

• Experience with agentic workflows (e.g., AutoGen, CrewAI).
• Familiarity with multi-modal models (text, image, etc.).
• Knowledge of advanced LLM fine-tuning techniques (e.g., LoRA, QLoRA).
• Strong SQL skills (especially with ClickHouse) and a keen eye for inference cost optimization.