Ishwarya G
Lead GenAI & ML Engineer| Agentic AI | Vertex AI Specialist |GCP ML Architect | Model Deployment & Monitoring | Feature Store | A/B Testing | Terraform | Python & Docker |Dataflow | ML Observability
- Role
- Lead Gen Ai Agentic Ai Engineer at GEICO
- Location
- Washington, DC, US
- LinkedIn followers
- 500 followers
About Ishwarya G
GenAI Engineer & Agentic AI Developer with 10+ years of experience in…
Experience
Lead Gen Ai Agentic Ai Engineer
Dec 2023 — Present · Chevy Chase, MD, US
Led the effort to build an Enterprise Data Lake on AWS Cloud, consolidating data from AWS and GCP to establish a single source of truth, enabling real-time analytics and AI/ML workloads.• Deployed LLM and Generative AI models using AWS Bedrock, Azure Open AI Studio, and Google Vertex AI, integrating them into enterprise workflows and applications.• Designed and developed Conversational AI solutions, integrating LLMs (GPT-4, BERT, T5) with Azure Open AI Studio, Microsoft Bot Framework (MBF) Bots, and CRM systems to enhance customer interaction.• Utilized Google Gemini/Gemini Pro to design scalable, high-performance GenAI architectures for large datasets and complex algorithms.• Implemented MCP-based agent architectures that dynamically route user prompts, tool outputs, and contextual metadata across tasks, improving agent coordination and reducing hallucinations by 35%.• Built agentic GenAI pipelines with A2A-inspired reflective loops, allowing models to critique and improve each other’s outputs in multi-agent RAG and Q&A systems.• Built custom instruction-tuned models with LLaMA-2 and QLoRA, using PEFT (Parameter-Efficient Fine-Tuning) and quantization-aware training to reduce memory usage by 60% without compromising accuracy in classification and generation tasks.• Designed and deployed scalable ML inference pipelines using Amazon SageMaker, leveraging built-in XGBoost and custom PyTorch containers for time-series forecasting models, reducing model deployment time by 50%.• Developed role-based access controls and security configurations in Microsoft Copilot Studio to ensure compliance with enterprise data governance policies and prevent data leakage during AI-driven interactions.• Architected a data lake on Amazon S3 to store structured and unstructured datasets (raw + processed), enabling serverless ETL workflows and ML training with seamless integration into SageMaker and Athena.
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.