AMIT YADAV
Lucknow
Summary
AI/ML Engineer and MLOps Lead with 8+ years of experience delivering production-grade ML and Generative AI systems across GCP, Azure, and AWS. Expert in MLOps pipelines (Kubeflow, Vertex AI, SageMaker, Azure ML), CI/CD automation, and scalable deployments on Kubernetes and Terraform. Collaborated with Google’s Agent Development Kit (ADK) within the Gemini team, building multi-agent systems for cybersecurity, healthcare, and energy domains. Skilled in LLM fine-tuning, Hugging Face models, RAG pipelines, and vector databases for domain-specific GenAI solutions. Integrated LangChain, OpenAI tools, Tavus Virtual Avatars, and observability platforms (Phoenix, W&B, AgentOps) into enterprise AI workflows. Proven team lead and individual contributor, delivering measurable outcomes across banking, telecom, manufacturing, and renewable energy.
Overview
8
8
years of professional experience
1
1
Certification
Work History
AI Software Designer –GenAI, Google Collaboration
Globant
10.2024 - Current
- Contributed in Google’s Agent Development Kit (ADK) as part of the Gemini/Generative AI team, enabling enterprise-scale agent frameworks and deployment of AI agents on Cloud Run and Vertex AI Agent Engine.
- Designed and implemented multi-agent systems for clients in cybersecurity, energy, and healthcare domains, evaluating agent behavior, accuracy, and latency under production workloads.
- Integrated LangChain tools, OpenAI functional tools, and custom toolchains into the ADK framework, expanding its interoperability and enabling richer agent workflows.
- Delivered a multimodal agentic system by integrating Tavus Virtual Avatar with ADK’s Live API for a major banking customer, supporting video, audio, and text interactions.
- Implemented observability and monitoring for multi-agent systems by integrating Arize Phoenix, Google Cloud native tracing, Weights & Biases (W&B), and AgentOps, ensuring transparency in function-calling, session tracking, and error analysis.
- Developed and deployed A2A protocol–based multi-agent architectures with MCP tools and Vertex AI memory bank for contextualized session memory and reliable orchestration.
- Acted as both team lead and individual contributor, guiding engineers, reviewing architecture, and delivering client-facing GenAI solutions in close collaboration with Google’s product and research teams.
Senior MIOps Engineer
Publicis Sapient Groupe
08.2023 - 08.2024
- Developed and fine-tuned Large Language Models (LLMs), including Hugging Face models, using LoRA and transfer learning to enhance domain-specific NLP applications.
- Implemented Retrieval-Augmented Generation (RAG) pipelines with vector databases (Milvus, Pinecone, PostgreSQL extensions) to enable semantic search and knowledge retrieval.
- Designed and automated MLOps pipelines using Kubeflow Pipelines across multi-cloud environments (GCP Vertex AI, AWS SageMaker, Azure ML Studio) for reproducible training, scalable deployment, and continuous monitoring.
- Led a team of MLOps/ML engineers, mentoring members, reviewing design/architecture, and driving project execution while also contributing as an individual contributor on complex hands-on development tasks.
- Built and orchestrated end-to-end CI/CD workflows with GitHub Actions, ensuring automated testing, versioning, and reliable delivery of ML services to production.
- Deployed and managed Kubernetes clusters on GCP GKE, AWS EKS, and Azure AKS, applying best practices for scaling, workload isolation, and governance.
- Developed and deployed microservices and APIs with FastAPI, secured and routed through APISIX, and integrated with cloud-native services for production-grade performance.
- Provisioned and managed infrastructure as code with Terraform, covering compute, networking, and managed ML services across multi-cloud.
- Containerized and versioned ML workloads using Google Container Registry (GCR), Azure Container Registry (ACR), and Amazon ECR, ensuring consistency across environments.
- Applied RBAC, network policies, and pod security policies to secure Kubernetes clusters, aligned with enterprise security and compliance standards.
- Set up monitoring and observability using Prometheus, Grafana, and ELK stack, integrated with Cloud Monitoring, CloudWatch, and Azure Monitor to track performance, latency, and anomalies.
- Collaborated with cross-functional teams and stakeholders to deliver scalable, production-ready AI/ML solutions with measurable business outcomes.
- Developed cost estimates and project specifications for proposals.
ML Engineer II- (MIOps-GCP)
General Mills
11.2022 - 08.2023
- Provisioned and managed GKE clusters to deploy Kubeflow components and build end-to-end MLOps pipelines.
- Containerized ML components with Docker for reproducibility and integrated them into Kubeflow Pipelines for scalable training and deployment.
- Utilized TensorFlow Extended (TFX) for pipeline orchestration in Vertex AI, ensuring seamless integration with Google Cloud services.
- Built automated CI/CD pipelines using GitHub Actions to streamline testing, container builds, and deployment workflows for ML services.
- Designed and deployed ML models as Cloud Functions, enabling lightweight, serverless inference at scale.
- Implemented demand forecasting pipelines leveraging BigQuery as the data warehouse for feature storage, training data, and analytics.
- Orchestrated data pipelines with GCP Composer (Airflow DAGs) to automate ETL, preprocessing, and model training workflows.
- Applied model monitoring and drift detection strategies to ensure ongoing reliability and accuracy of deployed models.
- Integrated Cloud Storage for dataset and model artifact management with high availability and scalability.
- Collaborated with cross-functional teams to deliver production-grade ML pipelines entirely on the GCP ecosystem.
Senior Data Scientist
Suzlon Group
03.2021 - 11.2022
- Led a team of data scientists and ML engineers in designing and delivering predictive maintenance solutions for wind turbines, while also contributing as an individual contributor on deep learning and deployment tasks.
- Designed and automated deep learning pipelines using 1D CNN models to detect anomalies in gearboxes and main bearing failures, deployed via Azure CI/CD pipelines with Docker for seamless production integration.
- Built data preprocessing workflows in Python and automated them with Azure Data Factory, ensuring reliable ETL for model training and inference.
- Developed and deployed a Computer Vision YOLO model for blade defect detection, integrated with drone imagery, and tracked experiments using MLflow for model logging and monitoring.
- Created predictive maintenance models leveraging machine learning and time-series forecasting to detect main bearing failures from oil sample and SCADA data.
- Applied advanced mathematical techniques (e.g., Fast Fourier Transform, Hilbert Transform) to analyze rotor vibrations in DFIG-based turbines for early fault detection.
- Built Suzlatics, a predictive analytics application for monitoring turbine main bearing temperatures, with visualization and reporting in Python.
- Utilized Azure Databricks and Snowflake for scalable data processing, integration, and analytics across the predictive maintenance ecosystem.
Machine learning Engineer
Thyssenkrupp Inc.
01.2020 - 02.2021
- Architected and deployed end-to-end AI/ML solutions on Azure Machine Learning, implementing automated ML pipelines for automotive parts demand forecasting (Volkswagen Portugal) using Facebook Prophet.
- Designed and implemented Azure DevOps CI/CD pipelines with YAML and the Azure ML SDK, automating model training, testing, and deployment workflows.
- Set up model monitoring and experiment tracking using Azure ML Studio and MLflow, enabling real-time performance monitoring, drift detection, and governance.
- Built and deployed Computer Vision models (YOLOv5) on Azure Kubernetes Service (AKS) for real-time defect detection in industrial conveyor systems, using Azure Container Registry for image management.
- Implemented Azure IoT Edge solutions for deploying SSD object detection models with OpenCV at the edge for steel plate quality control, leveraging Azure IoT Hub for device management.
- Developed and deployed LSTM models for time-series forecasting, served via Azure ML endpoints and integrated with Azure Synapse Analytics for downstream processing.
- Applied MLOps best practices including model versioning, A/B testing, and automated retraining using Azure ML pipelines and Azure Functions for event-driven model updates.
Data analyst
Convergys India Pvt. Ltd
05.2017 - 12.2019
- Collected, interpreted, and analyzed large-scale telecommunication datasets, delivering insights to support business decision-making and customer strategy.
- Performed predictive analytics (regression and classification) to forecast customer trends and improve retention strategies.
- Designed and implemented machine learning workflows (data processing, supervised and unsupervised learning) to enhance data analysis capabilities.
- Developed and deployed a conversational chatbot for customer support using NLP techniques, improving self-service query handling.
- Applied statistical sampling methods and developed custom statistical functions in Python to ensure accuracy and reliability in analysis.
- Built reporting dashboards and visualizations with Power BI and Tableau, supporting business leaders with real-time KPIs.
- Retrieved and processed large datasets using MySQL queries on AWS Redshift, integrating cloud data warehousing into analytical workflows.
Education
Executive MS - Artificial Intelligence
Aegis School of Business and Data Science
Powai, Mumbai01.2020
Skills
LLMs
Agentic framework(LangGraph, Adk)
Generative AI
MLOps
Vertex AI
Amazon Bedrock
OpenAI
Sagemaker
Github Action CI/CD
Kubeflow
Data Science
TensorFlow Extended
Kubernetes
Istio
Computer Vision
OpenCV
NLP
Deep Learning
Machine Learning
Signal Processing
Voice Analysis
APISIX
FastAPI
AWS
Azure
GCP
Docker
Prometheus
Grafana
ELK Stack
RBAC
Network Policies
ConfigMaps
Secrets
MTLS
Cloud Functions
Cloud Storage
Cloud Pub/Sub
Terraform
Demand Forecasting
Python
GCP Composer
Airflow
BigQuery
SQL
Flask
ML Flow
Azure DevOps
ETL
Azure Data Factory
Azure Databricks
Snowflake
Forecasting
AWS Forecasting
Time Series Analysis
Certification
Google Cloud Professional Machine Learning Engineer (Issued Feb 2025)
- Validates expertise in: ML model design, training, deployment, automation, and monitoring on GCP.
- Key Skills: Vertex AI, BigQuery ML, AutoML, ML APIs, TensorFlow Extended (TFX), MLflow integration, Responsible AI, model evaluation & retraining.
- Area of Expertise: End-to-end ML lifecycle on Google Cloud with production-grade automation and compliance.
Google Cloud Professional Cloud Architect (Issued June 2024)
- Validates expertise in: Designing, building, and securing scalable cloud infrastructure on GCP.
- Key Skills: GKE/Kubernetes, IAM & security, networking, storage systems, VPC design, hybrid/multi-cloud architecture.
- Area of Expertise: Architecting resilient, cost-optimized, and secure AI/ML and data platforms on GCP.
Google Cloud Professional Data Engineer (Certified – Year to update)
- Validates expertise in: Designing, building, operationalizing, and optimizing data pipelines and data processing systems.
- Key Skills: BigQuery, Dataflow, Dataproc, Pub/Sub, Cloud Storage, SQL optimization, ETL/ELT pipelines.
- Area of Expertise: Data engineering for AI/ML workflows — ingestion, transformation, warehousing, and real-time analytics.
Languages
English, Hindi and German(A1)
Timeline
AI Software Designer –GenAI, Google Collaboration
Globant
10.2024 - Current
Senior MIOps Engineer
Publicis Sapient Groupe
08.2023 - 08.2024
ML Engineer II- (MIOps-GCP)
General Mills
11.2022 - 08.2023
Senior Data Scientist
Suzlon Group
03.2021 - 11.2022
Machine learning Engineer
Thyssenkrupp Inc.
01.2020 - 02.2021
Data analyst
Convergys India Pvt. Ltd
05.2017 - 12.2019
Executive MS - Artificial Intelligence
Aegis School of Business and Data Science