Agentic AI that must be trusted
Tool orchestration, evaluation loops, memory boundaries, guardrails and production failure modes for systems where agents touch real workflows.
Independent AI & Quant Systems Architecture · Rome / Remote
I design AI and quantitative systems that survive production, markets and scrutiny.
AIgen Consult is the independent practice of Alessio Rocchi, AI Researcher & Systems Architect and Co-Founder / Head of Quant Research at AIgen Solutions. I help CTOs, fintech teams and AI-native companies turn agentic AI, reinforcement learning and systematic trading ideas into validated, observable production platforms.
- VMware Tanzu
- AIgen Solutions
- Neural Trading Group
- Aruba Cloud
- Stanford ML Specialization
- University of Milan CS
- OSCP
- GitHub blackms
Where I help
Useful when the hard part is not the demo. It is proof, integration and operation.
Quant research that needs validation
Walk-forward testing, reinforcement-learning workflows, regime analysis, capacity caveats, risk telemetry and honest robustness checks.
ML platforms that need engineering discipline
FastAPI services, PostgreSQL/TimescaleDB, queues, observability, Kubernetes, Docker and reproducible deployment for live ML workloads.
Architecture decisions under pressure
Tradeoff analysis, build-vs-buy choices, cloud-native architecture, security posture and implementation paths that teams can defend later.
Engagement modes
Plug in where the technical risk is highest.
2–4 weeks
AI / Quant Architecture Review
Turn an ambiguous AI or trading initiative into a defensible architecture, validation plan, delivery roadmap and cost envelope.
- System design and risk register
- Model/provider and infrastructure choices
- Validation strategy before scale-up
Focused delivery
Critical Build
Implement the hardest production components: agent backends, data ingestion, RL services, execution pipelines and platform glue.
- Production code and deployment pipeline
- Observability and operational runbooks
- Security and failure-mode hardening
Fractional ownership
Principal Architect
Senior technical ownership for teams that need architecture quality, delivery discipline and deep AI/quant/cloud judgement.
- Backlog shaping and technical direction
- Code review and architecture mentoring
- Vendor, infra and model governance
Evidence ledger
The claims are grounded in shipped systems, enterprise delivery and public technical work.
I keep the homepage proof-oriented on purpose: senior buyers do not need AI theatre. They need to know whether the person can validate, build and operate complex systems.
Staff / Principal Solution Architect · VMware Tanzu
Led AI/ML, Kubernetes and cloud-native initiatives; drove €60M+ in technical wins and trained 30+ field architects.
Co-Founder & Head of Quant Research · AIgen Solutions
Built intelligent trading infrastructure including a 9-microservice platform from data ingestion through live execution.
Solutions Architect · Aruba S.p.A.
Co-designed Aruba Cloud, automated provisioning from hours to minutes, built billing and metering for 6K+ tenants and supported €2M annual contracts.
Agentic software, external memory and algorithmic trading
Publications cover constrained LLM coding agents, external information management in LLM agents, hierarchical RL for trading and equity return prediction.
Complete profile signal
Research depth plus enterprise operating experience.
Co-Founder & Head of Quant Research · AIgen Solutions
Building intelligent trading infrastructure for fintech clients with measurable validation, live execution discipline and production engineering.
Founder & Chief Research Officer · Neural Trading Group
Independent quantitative research initiative focused on AI, machine learning, reinforcement learning and systematic trading.
Staff / Principal Solution Architect · VMware Tanzu
Drove AI/ML, Kubernetes and cloud-native initiatives, secured €60M+ in technical wins and trained 30+ field architects.
Solutions Architect · Aruba S.p.A.
Co-designed Aruba Cloud, automated provisioning from hours to minutes and supported billing/metering systems for 6K+ tenants.
github.com/blackms
Open-source work that maps to client delivery.
The public GitHub profile shows practical work across agent orchestration, Claude Code tooling, quant trading systems, cloud-native architecture and developer automation.
Open GitHub profile- 25
- public repositories
- 55
- public stars
- 52
- followers
aistack
Production-grade agent orchestration for Claude Code: 11 agents, 46 MCP tools, SQLite+FTS5, drift detection and consensus checkpoints.
SpreadPilot
Cloud-native options trading automation platform.
claude-flow-go
Go implementation of a multi-agent orchestration framework.
ExplainableEngine
Transforms numerical or decisional output into an explicit, queryable and verifiable causal chain.
Operating model
A small surface area, clear proof points and no unnecessary theatre.
-
01
Diagnose
Clarify business objective, uncertainty model, architecture reality, data constraints and operational risks.
-
02
Validate
Define walk-forward tests, robustness checks, negative controls, success metrics and failure modes before scaling.
-
03
Build
Implement the critical path with tests, automation, observability, deployment and operational runbooks.
-
04
Transfer
Leave the team with documented decisions, practical runbooks and a system they can safely extend.
PythonGoRustC#TypeScriptFastAPIPyTorchRay RLlibSAC/PPOLightGBMCVXPYVertex AI GeminiPostgreSQLTimescaleDBClickHouseKafkaRedisKubernetesDockerTerraformAnsibleArgoCDGitHub ActionsVMware TanzuObservability
Contact
Send the technical brief, not the pitch deck.
Include the objective, current stack, data reality, timeline and the part of the system that feels risky. I will reply with a concrete next step.
Independent freelance consulting under Italian P.IVA · Remote-first · Based in Rome, Italy