AI/Cloud consultant specializing in LLM inference infrastructure and MLOps. 5 years building production AI systems on AWS and Azure.
Quantitative equity strategy. Backtested 2015→2026: CAGR 21.1%, Sharpe 0.92. Deployed live via IBKR.
Fully automated forex trading system on 10 major pairs. ATR-based sizing, correlation filtering, stress regime detection. ~30% CAGR. Live track record building on Darwinex for external capital allocation.
Automated monitoring pipeline for quantitative crypto research. Production-grade Python package with GitHub Actions running every 4h: signal detection, paper trading engine, and weekly performance reporting — designed for unattended operation over a multi-month research period.
Sports Betting Quantitative Framework — live dashboard
Statistical model for identifying value bets in Premier League football markets. Detects mispricings in draw odds using historical match data, validated across 4 independent out-of-sample windows (Z = 3.37). +20.7% ROI, ×4.2 bankroll growth since Aug 2022.
Contributing to LLM inference and agentic frameworks:
Merged
- litellm#30537 - Add UK PII entity types to Presidio guardrail (merged <24h)
Active PRs
- strands-agents#2822 - Build-time validation for Bedrock strict_tools constraints (prevented 18h prod outage)
- strands-agents#2821 - Fix Anthropic ParsedTextBlock Pydantic serialization warnings
Focus areas: provider integrations (Anthropic, AWS Bedrock, Presidio), build-time validation, comprehensive testing with real APIs.
LLM & AI
- LLM inference optimization (litellm, strands-agents)
- Guardrails and content filtering (Presidio, AWS Bedrock)
- Model deployment and serving
Cloud & Infrastructure
- AWS (Bedrock, Lambda, ECS, SageMaker)
- Azure AI services
- MLOps pipelines
Languages & Tools
- Python, TypeScript
- Docker, Kubernetes
- Terraform, CloudFormation
- Ecole Centrale Casablanca / Ecole Centrale Marseille
- Based in Paris
- Open to consulting opportunities in LLM infrastructure
- Email: nbouyaakassinga@gmail.com
- LinkedIn: n-bouyaa-kassinga