Public Telegram Archive
Channels About
User Info
Eduard Babayan
ID 1513833915
Type user
First Name Eduard
Last Name Babayan
First Seen 2026-10-05 10:09 UTC
Updated 2026-10-05 10:09 UTC
Statistics
Total Messages: 1
Active in: 1 channel(s)
Activity by Channel
Channel Messages First Last
CY iT HR
@cyprusithr
1 2026-10-05 10:08 UTC 2026-10-05 10:08 UTC
Recent Messages
CY iT HR 2026-10-05 10:08 UTC
#CV #resume #ml #datascience #llm #rag #python # ai

Position: Senior Data Scientist / ML Engineer
Location: Yerevan, remote
Employment: full-time
Experience: 8+ years

About me:
Senior Data Scientist / ML Engineer working in regulated domains: oil and gas, healthcare, fintech and real estate. I'm currently building a multimodal LLM agent platform for an oil and gas project. It's a multi-agent system on Pydantic-AI and LangGraph with MCP, hybrid lexical + semantic search in OpenSearch, QLoRA fine-tuning, vLLM serving and async microservices on AKS. I added a cross-encoder re-ranking stage that lifted MRR by 15% with no added latency. I also built clinical AI for oncology: a HIPAA-compliant RAG pipeline over patient history and pathology reports, computer vision for pathology slides, and on-device Android models via ONNX Runtime. At Cognaize I worked on document intelligence and semantic search over financial and regulatory filings, using bi- and cross-encoders and learning-to-rank (LambdaMART). I also moved inference to ONNX Runtime + Triton, which cut latency by 60%.

Stack:
Python, PyTorch, TensorFlow, Hugging Face, vLLM, PEFT/QLoRA, LangChain, LangGraph, LlamaIndex, Pydantic-AI, MCP, RAGAS, MLflow, LangSmith, Scikit-learn, XGBoost, CatBoost, LightGBM, OpenCV, ONNX Runtime, NVIDIA Triton, FastAPI, asyncio, Neo4j, Elasticsearch, OpenSearch, PostgreSQL, Qdrant, FAISS, Docker, GitHub Actions, GitLab CI/CD, Azure (OpenAI, AI Search, AI Foundry, AKS), AWS (Bedrock, SageMaker, Lambda, ECS)

Languages: Armenian C2, Russian C2, English C1, German A2

Email: [email protected]