Agents alreadyrunning on Mondays.
Production AI / ML systems shipped before raiagents was founded, delivered while at gAI Ventures and Amazon, plus a solo-founded consumer app. Numbers are real and freshly measured. Reach out for unredacted references on request.
- C-001 · UNO Digital BankBanking · 2026
10-week production rollout of a 6-agent banking analytics platform.
LLM orchestration on AWS Bedrock that lets bank staff query a 10-table, 858-column datamart in plain English. Grounded with 262 golden queries and weighted few-shot retrieval. The first paying international client for the venture, shipped end-to-end with security review, audit trail, and zero data-mutation incidents.
Time-to-prod
10 wks
LLM latency
−40%
Mutation incidents
0
AWS BedrockClaudePostgresFastAPIMulti-agentFew-shot retrievalLead AI Product Engineer · gAI Ventures - C-002 · RoutesensePayments · 2026
Real-time multi-MID transaction routing, auth rates up 8–10%.
Built a real-time multi-MID transaction routing engine driven by MID health, approval rates and availability signals. Designed a MID health scoring system based on historical performance and risk indicators. Engineered a low-latency decision system with failover and adaptive load balancing. Built core payments infrastructure 0→prod.
Auth rate
+8–10%
Processing cost
−2–5%
Reliability
0→prod
Real-time decisioningMID scoringFailoverAdaptive LBDistributedFounding AI Engineer · Routesense - C-003 · Internal · SwikFintech · 2026
Invoice factoring SaaS: 12 AI services live in 6 weeks.
End-to-end build of Swik 2.0, 18-table Postgres schema, 90+ FastAPI endpoints, Next.js frontend. Replaced manual verification with an automated multi-channel engine: email playbooks, AI phone calls, vision-based portal automation, and a 3-stage PDF extraction chain. Verification compressed from days to hours.
Endpoints
90+
AI services
12
Verification
days→hrs
FastAPINext.jsPostgresPlaywrightClaudeVision LLMsLead AI Product Engineer · gAI Ventures - C-004 · Internal · SwikKnowledge / RAG · 2026
Real-time enterprise RAG across 5 fragmented SaaS sources.
Replaced overnight batch indexing with second-level data freshness. Push notifications, webhooks, and delta sync ingest from Gmail, Drive, Outlook, OneDrive and Dropbox. Per-user namespace isolation. Powers downstream agents with grounded retrieval and citation-level audit.
Sources
5
Freshness
~1s
Tenant isolation
Per-user
PineconeWebhooksDelta syncFastAPIPostgresOAuthLead AI Product Engineer · gAI Ventures - C-005 · Internal · SwikF&B / Risk ML · 2025
Predicting F&B operational distress 60–90 days before financials.
Distributed Scrapy + Playwright crawlers (proxy rotation) ingested 1.2M+ Yelp/Google reviews and health-inspection records across 40K+ California establishments. Time-series state-change detection (PELT/BOCPD) layered with a LightGBM classifier over 80+ features.
Top-decile precision
87%
ROC-AUC
0.82
FP cut
−55%
LightGBMPELTBOCPDScrapyPlaywrightTime-seriesML Engineer · gAI Ventures - C-006 · AmazonPayments · 2022–2025
Distributed payment ML safeguarding billions in daily transaction volume.
Event-driven payment pipelines that scaled to support ~30% of internal transactions, plus ML integrations for fraud and anomaly detection. Refactored backend services and streamlined deployment to cut infra costs while holding a 99.9% uptime SLA.
Payment risk
−78%
Inference latency
−45%
SLA
99.9%
LambdaSageMakerS3JavaEvent-drivenKinesisSoftware Engineer, ML · Amazon - C-007 · GamificationFitness LLCConsumer · Health · 2024–
Cerro: Apple Watch + LLM-agent fitness coaching with 70% beta retention lift.
Solo-founded fitness coaching app integrating Apple Watch and HealthKit with backend services and LLM-powered agents for personalized training and nutrition guidance. Gamified engagement loop and real-time progress tracking shaped retention.
Beta retention
+70%
Platform
iOS
Sensors
HealthKit
SwiftHealthKitFastAPIOpenAIPostgresFounder
Currently booking · 4 slots / mo
Tell us about your business.
We'll tell you what to automate first.
Free 30-minute discovery call. We talk through your business, where AI could actually help, what it would cost, and how long it would take. No pressure to commit. If we're not a fit, we'll say so.