Impetus - Machine Learning Engineer - Generative AI

Role : Machine Learning EngineerAbout Impetus : Impetus Technologies is a digital engineering company focused on delivering expert services and products to help enterprises achieve their transformation goals. We solve the analytics, AI, and cloud puzzle, enabling businesses to drive unmatched innovation and growth. Founded in 1991, we are cloud and data engineering leaders providing solutions to fortune 100 enterprises, headquartered in Los Gatos, California, with development centers in NOIDA, Indore, Gurugram, Bengaluru, Pune, and Hyderabad with over 3000 global team members.Skills : - Experience implementing supervised machine learning to real-world tabular data use cases.- Highly proficient in Python, Tensorflow/PyTorch and query languages like SQL/Hive/Pig; with the ability to use of app development tools such as Docker, Django, Git, etc.- Knowledge of machine learning algorithms like Classification, Regression, Clustering, Computer Vision, NLP/NLU techniques, etc.- Deep Learning architectures in Gen AI focused on LLMs (Transformers, CNNs, etc.)- Proficiency with open-source LLMs and tools such as GPTX.x, Anthropic Claude, Llama. SaaS LLMs like LangChain, llama index, Azure OpenAI, AWS Bedrock multimodal processing.- In-depth knowledge of Finetuning, Prompt Engineering, developing RAG pipelines.- AWS Services, including Bedrock, Sagemaker, Redshift, Athena, and S3 & AWS data lake.Key Responsibilities : - Architecture Design : Build enterprise Generative and Agentic AI platforms featuring high-performance RAG (Retrieval-Augmented Generation) pipelines and vector database integrations.- Agent Orchestration : Define multi-agent collaboration patterns, memory management, and autonomous planning frameworks.- Governance & Security : Implement data privacy, compliance, risk mitigation, and evaluation guardrails across all AI touchpoints.- Cross-functional Leadership : Guide and mentor engineering teams, run discovery workshops with stakeholders, and define reusable deployment patterns.Technical Stack : - Retrieval-Augmented Generation (RAG) pipelines, semantic caching, and context window optimization.- Function calling, tool use, and structured data extraction schemas.- Evaluation metrics, tracing, and hallucination reduction guardrails.- Designing agentic-first workflows and autonomous decision loops.- Multi-agent coordination patterns (supervisor-worker, decentralized collaboration, stateful graphs).- Frameworks like LangChain/LangGraph/Bedrock Core Runtime for state and memory management.- Emerging interoperability standards like Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocols.- Vector databases (e.g., Milvus, Amazon Aurora PostgreSQL) for high-speed similarity search.- Data pipelines and embedding generation workflows using Python, FastAPI, or Apache Spark.- Cloud-native deployment on platforms like AWS (Amazon Bedrock, Lambda, EKS, SageMaker, S3, RDS, DocumentDB)- Containerization and orchestration tools including Docker and Kubernetes.- CI/CD pipelines for automated testing of non-deterministic AI outputs.- Observability and logging pipelines for tracking agent token usage, latency, and failure states.- Responsible AI frameworks, data privacy compliance, and bias mitigation guardrails. (ref:hirist.tech)