AI Engineer | 4 Years in ML • 3 Years in GenAI/LLM | AWS Certified ML Engineer | MSc Warwick

I build production AI systems that actually work in the real world — not just notebooks and demos.

• 🎓 MSc Computer Science — University of Warwick (2024)
• 🏆 AWS Certified Machine Learning Engineer – Associate (MLA-C01)
• 🔭 Specialised in LangGraph multi-agent orchestration, RAG pipelines, and FastAPI backend development
• ☁️ All projects deployed on AWS EC2 with full observability (Prometheus/Grafana) and CI/CD
• 🌍 Based in London, UK | Open to AI Engineer & ML Engineer roles across UK and India

Production fintech ML system processing 31.9M financial transactions via PySpark on AWS. XGBoost + LightGBM ensemble achieving AUC-ROC 0.9857 and 81.23% recall on 6.3M test transactions. SHAP explainability for compliance auditing. React dashboard. Prometheus/Grafana monitoring. GitHub Actions CI/CD.

Python XGBoost LightGBM SHAP PySpark FastAPI React AWS EC2 Prometheus Grafana

Production RAG system retrieving evidence from 252,316 PubMed abstracts using Pinecone Vector Database retrieval via Claude Haiku. Faithfulness evaluation (Langfuse), improving answer reliability from ~0.45 to ~0.71. Evolved architecture from fine-tuned BioBART (LoRA/PEFT) to production RAG.

Python FAISS BM25 AWS Bedrock FastAPI React LLM RAG LoRA/PEFT

LangGraph multi-agent pipeline where 4 specialised agents (Parser, Guardrails, Processor, Aggregator) coordinate security audit workflows via MCP integrations. Human-in-the-loop approval gates. 100% success rate capturing hidden secrets in stress testing. <$0.01 per audit.

Python LangGraph LangChain MCP OpenAI Streamlit AWS EC2

Serverless analytics pipeline over NYC taxi trip data. S3 partitioned data lake, AWS Glue schema discovery, Athena SQL queries, Lambda automation — no always-on infrastructure.

Python AWS S3 AWS Glue Athena Lambda PySpark Parquet

Production API gateway with token-based rate limiting via Redis, Prometheus metrics, Grafana dashboards, and Streamlit admin dashboard. Full stack containerised with Docker Compose.

Python FastAPI Redis Prometheus Grafana Docker Streamlit

End-to-end credit risk prediction with automated GitHub Actions CI/CD — auto-deploys to AWS EC2 on every push to main.

Python scikit-learn XGBoost Docker GitHub Actions AWS EC2 Streamlit

AI/LLM: LangChain • LangGraph • RAG • FAISS • BM25 • Prompt Engineering • LoRA/PEFT • OpenAI API • AWS Bedrock • SHAP

ML/Data: PyTorch • TensorFlow • XGBoost • LightGBM • scikit-learn • PySpark • Pandas • NumPy

Backend: Python • FastAPI • REST APIs • asyncio • Redis • Docker • Nginx • PostgreSQL

Cloud/DevOps: AWS (EC2 • S3 • Lambda • Glue • Athena • Bedrock • CloudWatch) • Terraform • GitHub Actions • Prometheus • Grafana

Frontend: React • TypeScript • Streamlit • Power BI

I'm actively looking for AI Engineer and ML Engineer roles in the UK and India.

• 🌐 Portfolio: vatsalsangani.in
• 💼 LinkedIn: linkedin.com/in/vats-sangani
• 📧 Email: vatsalsangani24@gmail.com