What I do

Core expertise

Deep specialization in AI engineering across multiple regulated and high-stakes industries — pharma, banking, audit, retail, and transport.

01

Agentic AI Systems

↓ 1–2 days → under 5 min (ICF generation)

Multi-agent architectures with scoped tool restrictions, Skills Cards patterns, and MCP integrations — automating complex multi-step workflows in enterprise environments.

DSPyLangGraphMCPSkillCards
02

RAG & Vision RAG

↑ Retrieval hit-rate@6: 60% → 85%

Advanced retrieval pipelines for text and multimodal content — hybrid search, reranking, decomposition-based retrieval, and vision processing for complex documents with tables and figures.

WeaviateVespaColPaliDoclingReranking
03

LLM Optimization & Eval

↑ Answer accuracy: 70% → 94% (DSPy optimizers)

Cost-aware LLM engineering with comprehensive evaluation frameworks — parsing quality, retriever performance, anti-hallucination robustness, and answer generation effectiveness.

LangFuseDSPy OptimizersEval PipelinesCost Tracking
04

Regulated & Domain-Specific AI

AI systems built for industries where compliance, precision, and traceability matter — pharma (ICF generation, medical writing), banking (KYC, fraud), audit (document intelligence), retail (recommendation engines), and transport (customer support).

PharmaBankingAuditRetailTransport
05

Quantitative Foundations

PhD in Applied Mathematics — stochastic differential equations, Lévy processes, Malliavin calculus. Years of pricing models and risk analytics at Société Générale, Thomson-Reuters, and RBC Dexia.

Stochastic ModelsPortfolio OptimizationPricingRisk
06

Full-Stack AI Shipping

↓ Response time: ~2 min → ~4 sec (streaming)

From FastAPI backends with full streaming architecture to Docker/K8s deployments and MLOps pipelines — shipping production-grade AI applications end-to-end on AWS, Azure, and GCP.

FastAPIDockerK8sSageMakerMLFlow