Skip to content

Nikhil Ramavath

AI/ML Engineer • St. Louis, Missouri • r***************@gmail.com • +13******246 • linkedin_nikhil/••••• • drivetube.ai/•••••

Professional Summary

AI/ML Engineer with 5+ years architecting production-grade Generative AI, Deep Learning, and MLOps solutions for healthcare, pharma, and enterprise clients. Proven track record building LLM fine-tuning (LoRA, PEFT), RAG, and agentic systems (GPT-4o, Claude 3.5, Llama 3) and deploying scalable inference on Kubernetes and cloud ML platforms. Deep experience with BioNLP models (BioBERT, ClinicalBERT), PyTorch, PySpark, MLflow, and CI/CD for models; delivered multi-million dollar savings, reduced operational latency, and maintained HIPAA/GxP-compliant ML workflows.

Technical Skills

Programming Languages: Python,R
Frameworks and Libraries: Pandas,NumPy,PyTorch,TensorFlow,Keras,Scikit-learn,XGBoost,LightGBM,CatBoost,LangChain,OpenCV,FastAPI,Flask
Databases: SQL,Snowflake,PostgreSQL,MongoDB,Redis,Elasticsearch
Cloud and DevOps: MLflow,Kubeflow,Docker,Kubernetes,EKS,AKS,CI,CD,GitHub Actions,Terraform,AWS SageMaker,AWS EC2,S3,AWS Lambda,Azure Machine Learning,Azure Databricks,GCP Vertex AI
Data and Analytics: Apache Spark,Kafka,Apache Airflow,dbt
Skills: PySpark,Dask,Polars
LLM & Generative AI: Hugging Face Transformers,OpenAI API,Anthropic API,GPT-4o,Claude 3.5,Llama 3,Fine-tuning LoRA,PEFT,Prompt Engineering,Retrieval-Augmented Generation
NLP & BioNLP: BERT,BioBERT,ClinicalBERT,SciBERT,spaCy,NLTK,NER,Text Classification,Question Answering,Summarization
Computer Vision & Multi-modal: ResNet,EfficientNet,YOLO,U-Net,Albumentations
Vector DBs & Search: Pinecone,Weaviate,Milvus,ChromaDB,Qdrant,FAISS
Experimentation & Optimization: Optuna,Ray Tune,Weights & Biases,TensorBoard,Neptune,Hyperopt,Ray
Explainability & Monitoring: SHAP,LIME,Arize AI,Evidently AI,Great Expectations
APIs & Fast Prototyping: RESTful APIs,Streamlit,Gradio

Work Experience

Blue Cross Blue Shield Association
Chicago, IL
AI/ML Engineer
Sep 2025 – Present
Health insurance organization; built generative-AI and ML systems to automate claims, prior authorization, coding, and member risk-scoring for millions of members.
Tech Stack: GPT-4o, Claude 3.5, LangChain, Pinecone, BioBERT, ClinicalBERT, PyTorch, PySpark, AWS SageMaker, MLflow, Kubernetes, Presidio, Arize AI, Evidently AI
  • Architected and deployed an LLM-powered medical claims processing pipeline using GPT-4o and Claude 3.5 with LangChain that reduced manual review time by 78% and generated $12M in annual savings.
  • Built a retrieval-augmented generation system (Pinecone, LangChain) indexing 10M medical guidelines to automate prior authorization decisions, achieving 96% decision accuracy and faster approvals.
  • Developed medical coding automation by fine-tuning BioBERT/ClinicalBERT to assign ICD-10 and CPT codes, achieving 92% accuracy and reducing coding backlogs by 85%.
  • Engineered patient risk stratification models for 8M+ members using PyTorch and PySpark, delivering an 89% AUC-ROC for hospital readmission prediction used in care management workflows.
  • Fine-tuned Llama 3 using PEFT/LoRA on de-identified clinical datasets and implemented a Presidio-based PHI masking layer to remove PHI before external calls, cutting external API spend by 35% while preserving compliance.
  • Implemented end-to-end MLOps pipelines (AWS SageMaker, MLflow, Kubernetes) and model monitoring (Arize, Evidently) to automate versioning, CI/CD, and drift detection, maintaining 95%+ production accuracy and reducing ops time.
Abbvie
Vernon Hills, IL
AI/ML Engineer
Aug 2023 – Aug 2025
Pharmaceutical research and development; built ML systems for drug discovery, clinical trial analysis, and regulatory-compliant model interpretability.
Tech Stack: PyTorch, PyTorch Geometric, SciBERT, BioBERT, Hugging Face, Azure Databricks, MLflow, Great Expectations, PySpark, AKS, Docker
  • Designed a drug-discovery screening pipeline with Graph Neural Networks using PyTorch Geometric to evaluate 2M+ compounds, cutting R&D screening costs by $8M and accelerating candidate selection by 60%.
  • Built a literature mining system using SciBERT and BioBERT with Hugging Face to extract drug–disease relationships from 5M+ PubMed articles, enabling automated hypothesis generation for medicinal chemistry.
  • Developed adverse event prediction models on clinical trial datasets using PyTorch, improving early patient-safety signal detection with 88% prediction accuracy.
  • Orchestrated a centralized ML platform on Azure Databricks with MLflow, tracking 500+ model iterations and implementing Great Expectations for automated clinical data quality validation to support GxP workflows.
  • Engineered distributed PySpark pipelines to transform terabyte-scale genomic and clinical data, improving ETL throughput by 4x and shortening model training prep time.
  • Packaged and deployed predictive models as containerized REST APIs on AKS and optimized training with Mixed Precision and Distributed Data Parallelism, reducing training time from 5 days to 18 hours.
Quanteon Solutions
Hyderabad, India
Machine Learning Engineer
Jun 2020 – Jul 2023
Technology consulting and solutions provider delivering ML systems for telecom, e-commerce, fintech and manufacturing clients.
Tech Stack: XGBoost, CatBoost, Isolation Forest, Autoencoders, FastAPI, Docker, AWS ECS, Apache Airflow, Featuretools, PySpark
  • Developed a customer churn prediction model for a telecom client using XGBoost and CatBoost that achieved 91% AUC-ROC and enabled interventions saving $4.5M annually.
  • Built an e-commerce recommendation engine employing collaborative filtering and Neural Collaborative Filtering that increased CTR by 52% and generated $12M in incremental revenue.
  • Implemented a fraud-detection platform for a fintech client using Isolation Forest and Autoencoders, processing 10M+ daily transactions and preventing $8M in fraudulent losses with 94% precision.
  • Built a high-performance model-serving stack (FastAPI, Docker, AWS ECS) delivering 20K+ predictions per second with P95 latency under 80ms to support real-time decisioning.
  • Developed automated ETL and feature engineering pipelines with Apache Airflow and Featuretools processing 500GB+ daily; also implemented nightly PySpark batch jobs processing 50M+ records, reducing runtime from 8 hours to 1.5 hours.
  • Established an experimentation and A/B testing framework using Bayesian methods to evaluate 30+ model variants, measuring impact on conversion and retention for multiple clients.

Education

Webster University
Master of Science in Data Analytics • St. Louis, MO • Aug 2023 – Dec 2025
MVSR Engineering College
Bachelor of Technology in Electrical Engineering • Hyderabad, India • Aug 2017 – May 2021

Powered by Drivetube · Create your own profile at drivetube.ai