AI/ML Engineer

Job Title: AI/ML Engineer Location: 100% Work from home (Work in PST) Duration: 6 Months Contract with Potential Extension Must have Skills: Technical tools/applications would like the candidates to be proficient • Minimum years of experience: 5 years • AI/ML, LLM, Python, Agentic AI • Python, SQL, Snowflake Job Description: • Client is seeking an AI/ML Engineer to build enterprise-grade chatbots and intelligent assistants using state-of-the-art LLM technologies. • The role focuses on finance data, RAG-based architectures, hallucination mitigation, and agentic AI systems deployed at scale. • You will work closely with Product, Data, and Platform teams to deliver reliable, explainable, and production-ready AI solutions. Project information: • Developing AI-driven financial chatbot is mostly built and live in production. Designed to reduce repetitive querying by FPPS teams and financial analysts. • Supports financial reporting, forecasting, revenue recognition, and ACV reporting. Improves analyst efficiency and decision-making by automating access to financial data. Responsibilities: • Build LLM-powered chatbots using OpenAI, RAG, tool calling, and agent frameworks (LangChain, LangGraph) • Design Agent-to-Agent (A2A) architectures for multi-step reasoning and autonomous workflows • Design retrieval pipelines using vector databases • Implement hallucination reduction techniques: grounding, re-ranking, citations, confidence scoring • Work with finance and enterprise datasets ensuring accuracy and governance • Deploy and monitor AI systems using cloud-native and MLOps practices • Implement CI/CD for AI pipelines and inference services Technologies & Skills: • Python, SQL • LLMs: OpenAI (GPT-4/4.1), Anthropic, Gemini, Llama • Agentic AI: LangGraph, LangChain, Agent-to-Agent (A2A) patterns • RAG & Search: embeddings, hybrid search, cross-encoders • Vector Databases • Evaluation & Observability: LangSmith, MLflow, Weights & Biases • Cloud: AWS (S3, Lambda, SageMaker, Bedrock) • Data: Snowflake, DBT, structured & unstructured data pipelines • Evaluation: prompt/version management, offline & online LLM evaluation