Krishna Chaitanya Valaboju

AI Engineer at Longtermhealth GmbH

At an early-stage healthcare startup, built a full AI pipeline in Python using OpenAI API and OpenCV on AWS Fargate/ECS and reduced API calls by 30% while maintaining 99.9% uptime for real-time health analytics. Containerized ML workflows for sleep-phase analysis using MLflow and Docker, training multiple models achieving ROC-AUC of 0.87. Deployed a sub-100ms inference API using FastAPI and Docker, resulting in 20% increased mobile engagement and implemented CI/CD pipelines for zero-downtime deployments and 50% faster releases.

Computer Vision Engineer at BAUER Maschinen GmbH

Implemented real-time defect detection pipelines in PyTorch (YOLOv8 & Mask R-CNN) reducing inference latency by 25% and improving accuracy. Processed 3D point clouds with Open3D and applied voxel downsampling and RANSAC-based segmentation improving edge-case classification accuracy by 12%. Developed interactive dashboards with Dash & Matplotlib that enhanced model troubleshooting efficiency by 30%. Containerized and deployed vision services using Docker Compose and Kubernetes in ROS test rigs for scalable edge performance.

Machine Learning Engineer at Infralytica GmbH

Architected a hybrid AI pipeline combining regex, BERT embeddings, logistic regression, and Llama 3.0 inference, cutting costs by 30% and delivering a 92% F1-score for log classification. Developed crack detection using PyTorch U-Net and OpenCV with data augmentation and transfer learning achieving 0.75 IoU and 0.78 F1-score on unseen data. Built scalable MLOps workflows on Azure with MLflow, FastAPI, Docker, and Kubernetes, implementing CI/CD with GitHub Actions and monitoring with Grafana and AWS CloudWatch.

Machine Learning Engineer at Tata Consultancy Services

Developed a claims prediction model in Python utilizing Pandas, scikit-learn, XGBoost, and Random Forest improving approval accuracy by 15% and ROC-AUC to 0.82. Managed data extraction, cleaning, feature engineering, dimensionality reduction, and imbalance handling. Containerized ML services with Docker and Kubernetes deployed on AWS with automated CI/CD pipelines and implemented real-time model monitoring with Prometheus and Grafana. Awarded On The Spot Award and Fresco Play Miles Award for performance.

Generative AI Engineer at Long termhealth GmbH

Developed an automated podcast tool using LLM-based topic extraction and cosine similarity thresholding, boosting F1 score from 0.53 to 0.72 over the TextTiling baseline. Created and released a 30-episode hierarchically annotated gold corpus and performed rigorous leave-one-out cross-validation to benchmark segmentation methods on noisy ASR data.

Machine Learning Engineer at Infralytica GmbH

Built a full end-to-end Retrieval-Augmented Generation (RAG) pipeline in Python using OpenAI Assistants (GPT-4o), OpenAI and Hugging Face embeddings, ChromaDB, and OpenCV on AWS Cloud. Reduced API call costs by 30% and improved retrieval with Maximal Marginal Relevance (MMR), cross-encoder reranking, and query expansion. Developed a modular blog-generation Agentic AI application using FastAPI, Pydantic, LangChain, Groq LLM (LLaMA 3.1, 8B), and LangGraph Studio (LangSmith) for orchestration with support for multi-language translation and environment-based configuration and monitoring. Created a sleep quality detection model integrating Apple Watch data using supervised and unsupervised models (XGBoost, Random Forest, K-Means, DBSCAN, PCA), achieving an ROC-AUC of 0.87.

Computer Vision Engineer at BAUER Maschinen GmbH

Architected a hybrid AI pipeline combining regex pattern matching, BERT embeddings with logistic regression, and Llama 3.0 inference via Groq Cloud, reducing processing costs by 30% and delivering a 92% F1 score for log classification. Developed a crack detection system using PyTorch U-Net and OpenCV for industrial image segmentation; applied data augmentation and transfer learning achieving 0.75 IoU and 0.78 F1 on unseen crack datasets. Built scalable MLOps workflows with MLflow, FastAPI, Docker, and Kubernetes on GCP cloud, implemented CI/CD with GitHub Actions, and set up monitoring with Grafana and AWS CloudWatch.

Machine Learning Engineer at Tata Consultancy Services

Built a health insurance claims prediction model using Python (Pandas, scikit-learn, XGBoost, Random Forest), improving approval accuracy by 15% and raising ROC-AUC to 0.82. Extracted and cleaned data using SQL (PostgreSQL/MySQL) and engineered features, applying PCA for dimensionality reduction and SMOTE for imbalance handling. Tuned hyperparameters with Optuna and validated models with GridSearchCV to ensure robust performance. Containerized ML services with Docker and Kubernetes, deployed on AWS SageMaker and ECS/Fargate. Automated CI/CD pipelines using GitHub Actions and managed experiment tracking and model versioning with MLflow. Set up real-time monitoring with Prometheus and Grafana. Received On The Spot Award and Fresco Play Miles Award. Published at ACL '23 and secured 1st place in ImageArg '23.

Data Scientist at Tata Consultancy Services

Built a health insurance claims prediction model in Python using Pandas, scikit-learn, XGBoost, and Random Forest, achieving +15% accuracy and ROC-AUC 0.82. Data extraction/cleaning with SQL (PostgreSQL/MySQL); feature engineering; dimensionality reduction; hyperparameter tuning (GridSearchCV, Optuna). Containerized ML services with Docker and Kubernetes; deployed on AWS SageMaker/ECS/Fargate; automated CI/CD with GitHub Actions; MLflow for experiment tracking; Prometheus/ Grafana monitoring.

AI Engineer at Bluesphere.health

Led end-to-end RAG pipeline in Python using OpenAI GPT-4o, OpenAI Embeddings, Hugging Face embeddings, and ChromaDB on AWS. Annotated image data with CVAT; improved retrieval with MMR and cross-encoder reranking; contributed to data engineering and cloud deployment.