A selection of production-grade AI systems and distributed platforms I've designed and shipped.
AI ENGINEERING
Autonomous agent that scans repositories, identifies CVEs and security anti-patterns, and auto-generates patched PRs with explanations.
AI ENGINEERING
Context-aware code completion trained on internal codebases — repo-aware suggestions beyond stock Copilot.
AI ENGINEERING
Reads production code plus specs and generates unit, integration, and edge-case tests — lifted pilot coverage from 62% to 91%.
AI ENGINEERING
Converts Confluence / Jira specs into scaffolded services, DB schemas, and API contracts to accelerate greenfield delivery.
AI ENGINEERING
PR bot that reviews for correctness, security, and style — leaves inline comments with suggested diffs.
AI ENGINEERING
Ingests logs, metrics, and traces, correlates anomalies, and produces root-cause hypotheses during live incidents.
AI ENGINEERING
Chat over internal wikis, runbooks, and code — grounded answers with citations and access-control-aware retrieval.
AI ENGINEERING
Automates migration of legacy APIs (SOAP → REST / GraphQL) with contract validation and regression tests.
AI ENGINEERING
Classifies incoming tickets, drafts responses, and resolves L1 issues autonomously — deflected ~40% of tier-1 volume.
AI ENGINEERING
Monitors data pipelines, detects schema drift and anomalies, suggests remediations for data engineering teams.
AI ENGINEERING
Ingests meeting transcripts, extracts decisions and action items, assigns owners, and syncs to Jira / Linear.
AI ENGINEERING
Turns code, PR history, and commits into always-current architecture docs, ADRs, and onboarding guides.
AI ENGINEERING
Runtime guard inspecting LLM inputs and outputs for injection attacks, jailbreaks, and PII leakage.
AI ENGINEERING
Search a product catalog with text and images, reranked by user context for e-commerce and support use cases.
AI ENGINEERING
Auto-routes prompts between models based on task complexity, budget, and SLA — cut monthly LLM spend by ~45%.
DISTRIBUTED SYSTEMS
Microservices for claims, payments, notifications, and customer management. Saga pattern for transaction consistency; 10M+ events/day on Kafka; 99.99% availability.
DISTRIBUTED SYSTEMS
High-throughput pipeline with sub-100ms end-to-end latency, proven at 100k+ events/sec.
DISTRIBUTED SYSTEMS
Lift-and-reshape of legacy platform to AWS EKS with multi-AZ HA. IaC + GitOps cut infra costs ~35%.
DISTRIBUTED SYSTEMS
Per-tenant isolation, quotas, and near-real-time dashboards for a SaaS platform serving 500+ customers.
DISTRIBUTED SYSTEMS
Redis-backed sliding-window rate limiter with per-tier quotas, degradation modes, and single-digit-ms overhead.
DISTRIBUTED SYSTEMS
Immutable event store for compliance-grade auditability — SOC2 / HIPAA-ready, tamper-evident, replayable.
DISTRIBUTED SYSTEMS
Dual-write, backfill, and reconciliation tooling to migrate live schemas and databases without maintenance windows.
PLATFORM ENGINEERING
Backstage-based portal unifying service catalog, docs, on-call, and self-service infra provisioning for 100+ engineers.
PLATFORM ENGINEERING
SDK + control plane for progressive rollouts, killswitches, and experiment targeting — sub-10ms flag evaluation.
PLATFORM ENGINEERING
Custom gateway handling mTLS, OAuth2, rate limiting, request signing, and full-request tracing.
PLATFORM ENGINEERING
CLI + templates for a golden-path service — pre-wired observability, auth, CI/CD, and SLOs from day one.