Building
Engineered for scale.
I build cloud-native, event-driven systems and internal platforms with a focus on observability, performance, and production reliability.
InferGrid
A distributed ML inference platform built to handle production traffic, not just serve predictions. Requests are routed inline or to a Kafka-backed worker pool depending on in-flight load, with results delivered over WebSocket. A drift detector runs a KS-test against a baseline confidence distribution every 60 seconds to catch silent model degradation, and an Experiment Tracker service backed by PostgreSQL drives weighted A/B traffic splits with timeout-based fallback. Load tested to 200 concurrent users with zero failures across 67,963 requests — 699 RPS on the async path, sub-100ms p95 on the sync path up to 50 users.
699 RPS · 0 failures @ 200 usersVectorForge
A vector search engine written from the ground up — the HNSW index, the on-disk format, and the query layer are all hand-built rather than wrapped around an existing library. A multi-layer Hierarchical Navigable Small World graph handles approximate nearest-neighbour search, with an exact brute-force index alongside it as ground truth so every recall number is measured, not assumed. It runs single-node or as a sharded cluster: a from-scratch consistent hash ring routes writes by vector id, and a coordinator fans searches across shards and merges the top-k. Adds metadata filtering during traversal, a versioned binary persistence format, REST and gRPC surfaces, and Prometheus-backed recall-versus-latency dashboards.
0.97 recall@10 · 7.8ms p95Nexus
A comprehensive distributed observability platform that provides monitoring, logging, and visualization capabilities for complex systems.
Observability PlatformPulseForge
A robust event-driven backend platform designed for high-performance, scalable applications with asynchronous processing.
Backend PlatformMini ML Platform
End-to-end training + inference with MLflow registry
ML PlatformProcuroid
Autonomous multi-agent procurement platform (AI ATL 2025 HM)
AI ATL 2025 Honorable MentionCrumb
Realtime social app with NFC-based friend adding
Social App- Java (Spring Boot)
- Python (Flask, FastAPI)
- REST APIs
- Asynchronous Processing
- JWT & Role-Based Access Control
- System Design
- AWS (EC2, S3, Lambda, IAM, CloudWatch)
- Google Cloud Platform
- Docker
- Kubernetes (Kind / Minikube / EKS)
- CI/CD Pipelines
- Infrastructure Cost Optimization
- Prometheus
- Grafana
- ELK Stack
- Service Level Objectives (SLOs)
- Error Budgets
- Monitoring & Alerting
- scikit-learn
- MLflow (Tracking & Registry)
- Model Training Pipelines
- Production Inference APIs
- Experiment Versioning
- React
- Next.js
- TypeScript
- Tailwind CSS
- Browser Extensions
- Responsive UI Design
- Swift (iOS)
- Kotlin (Android)
- MVVM Architecture
- React Native
- Expo
- Mobile REST Integrations
Platform automation, internal tooling, reliability systems development and maintenance.
Cloud services development, microservices architecture, observability implementation, and Kubernetes deployment.
ML systems development, graph models research, AWS serverless architecture, and data engineering.
University of Georgia
Let's build something.
Open to full-time roles, contracts, and interesting problems in backend, cloud, or AI infra.