Milan Spasov
ML Engineer
12+ years ML exp
Python
SQL
Active 8 days ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
Senior AI/ML engineer with 10+ years of experience in machine learning, AI research, and deployment. Builds and deploys LLM/NLP and multimodal systems with RAG, knowledge graphs, and RLHF, and supports production delivery with MLOps and CI/CD. Experience spans healthcare and finance use cases, plus research contributions in deep reinforcement learning at Google DeepMind.
Technical skills
Python
SQL
Python
Django
FastAPI
Databases
Apache Kafka
Google BigQuery
Pinecone
Weaviate
FAISS
Neo4j
AI/ML
Airflow
BERT
Claude
Computer Vision
Explainable AI
Fine-tuning
Hallucination
KServe
Llama
LlamaIndex
Model Context Protocol
Multimodal AI
ONNX
Prefect
Prompt Engineering
Ray Serve
Reranking
Sentiment Analysis
Spark
TensorRT
vLLM
Ray
Scikit-learn• 12y+
XGBoost• 12y+
TensorFlow• 11y+
LIME• 10y+
PyTorch• 10y+
Reinforcement Learning• 9y+
spaCy• 6y+
Keras• 3y+
Kubeflow• 3y+
LangChain• 3y+
LightGBM• 3y+
OpenAI Evals• 3y+
SHAP• 3y+
AWS Bedrock
LangGraph
LangSmith
RAG
RLHF
TensorRT-LLM
Diffusion Models
LLM
NER
NLP
Self-Supervised Learning
Transfer Learning
Transformers
DevOps
Git
GitHub Actions
Helm
Jenkins
Prometheus
Rest API
AWS• 12y+
Docker• 12y+
Kubernetes• 3y+
Terraform• 3y+
Amazon EKS
CI/CD
Azure
GCP
Robotics
Reinforcement Learning• 6y+
Timeline
Senior AI/ML Engineer
•
Senior
Sword Health
•
Full-Time
Led development of a multi-agent conversational AI system for patient recovery support, using LangChain/LangGraph and domain adaptation over patient data. Built a hybrid RAG pipeline with Pinecone and FAISS plus re-ranking, and delivered graph-based reasoning with Neo4j for multimodal clinical inputs. Implemented an RLHF optimization pipeline and deployed scalable MLOps on AWS Bedrock and EKS with optimized inference to improve throughput and latency, alongside CI/CD and LLM evaluation monitoring.
LangChain
LangGraph
RAG
FAISS
Neo4j
RLHF
AWS Bedrock
Amazon EKS
TensorRT-LLM
LangSmith
OpenAI Evals
CI/CD
Staff Machine Learning Engineer
•
Lead
Stripe
•
Full-Time
Developed fraud detection models combining XGBoost/LightGBM and deep learning to reduce false positives in financial transactions. Built a real-time prediction system using TensorFlow/Keras and delivered interpretability workflows with SHAP and LIME for explainable fraud decisions. Integrated LangChain for personalized recommendations, and implemented production ML pipelines with Kubeflow, Kubernetes, and Terraform focused on scalable model deployment and management, plus research into multi-agent reinforcement learning for payment decisioning.
XGBoost
LightGBM
TensorFlow
Keras
LangChainsince 2023
Kubeflow
Kubernetes
Terraform
SHAP
LIME
Reinforcement Learning
Senior Data Scientist / ML Engineer
•
Senior
SAP
•
Full-Time
Developed and deployed predictive time-series models for enterprise supply chain optimization, including LSTM/GRU-style forecasting approaches. Built NLP pipelines with spaCy and Hugging Face for document classification and information extraction from unstructured enterprise data. Implemented deep learning solutions with TensorFlow and PyTorch and set up cloud-based MLOps using SAP Cloud Platform and AWS to support continuous training and deployment across business units.
spaCy
TensorFlow
PyTorch
AWS
AI Research Engineer
•
Middle
Google DeepMind
•
Full-Time
Conducted research in deep reinforcement learning and neural network methods for real-world agent and robotics-style control tasks. Worked on meta-learning and transfer learning approaches to improve learning efficiency across environments, including contributions toward AlphaZero-related architectures. Participated in neural architecture search work to reduce model complexity while maintaining performance, focusing on scaling reinforcement learning training efficiency and compute cost.
Reinforcement Learningsince 2017
Junior AI/ML Developer
•
Junior
Scalefocus
•
Full-Time
Built predictive models for enterprise clients using scikit-learn and XGBoost to support automated insights and business decision-making. Developed AI-powered forecasting tools to improve trend prediction accuracy and supported NLP model development for extracting key information from business documents. Gained hands-on deployment experience with Docker and AWS, enabling scalable model hosting and reliable availability.
Scikit-learn
XGBoostsince 2014
Docker
AWSsince 2014
TensorFlowsince 2014
PyTorchsince 2014
Stanford University
Master's Degree •
Computer Science