Sainitheesh Erra
Machine Learning Engineer and GenAI, Agentic AI | LLM | Data Engineering | Predictive Analytics | VertexAI | Azure | AWS | ML | NLP | Big Data
- Role
- Machine Learning Engineer (Generative Ai) at Amgen
- Location
- Atlanta, GA, US
- LinkedIn followers
- 500 followers
About Sainitheesh Erra
Machine Learning Engineer with over 4+ years of experience in building Machine Learning…
Experience
Machine Learning Engineer (Generative Ai)
May 2024 — Present · Atlanta, GA, US
Built a document-chatbot RAG backend on GCP, ingested PDFs from Cloud Storage, chunked & embedded them with OpenAI text-embedding models, and served vectors via Redis, pgvector, chromaDB for low-latency retrieval.• Orchestrated the pipeline with Vertex AI Pipelines (data prep → embedding → index update → eval), replacing ad-hoc scripts; leveraged Cloud Functions, Cloud Run for event-driven index refreshes on new file batch drops.• Added agentic workflows using Vertex AI Agents, Google ADK and LangGraph agents for search, summarization, routing and citation verification, enabling multi-step reasoning (e.g, auto-follow-up queries, policy lookup) and autonomous report drafting.• Implemented an evaluation of 2 layers with Guardrails + custom BLEU, ROUGE, faithfulness checks in first layer and then followed by LLM as judge in second layer, scheduled CI tests to gate model,prompt changes before promotion.• Developed a complete matured MLops template end-end CI/CD architecture which is used for multiple machine Learning models throughout the team that includes feature engineering, model training, using model registry, pub/sub, model deployment using Kubeflow pipelines in vertex AI, worked on gathering requirements, developed model code using MLops structure which I developed, deployed the models into production using own architecture that I developed for all models in Vertex AI• Exposed a FastAPI/Cloud Run endpoint for real-time Q&A and summary generation, with IAM and CMEK encryption for PHI/PII, and fine-grained access via VPC-SC; added prompt templates and per-tenant guardrails for safe, compliantresponses.• Experimented with GPT-J, GPT-NeoX, and T5 LLMs, Reinforcement Learning with Human Feedback, and Direct Preference Optimization with Parameter Efficient Fine-Tuning and DeepSpeed to construct a cost-effective customer representative chatbot in a limited GPU setting.
Education
CVR College of Engineering, Hyderabad
Bachelor of Technology - BTech, Computer Science
Georgia State University
Master of Science - MS, Data Science
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.