// Operations

Field Proven Systems

These are the systems behind the strategy.

The work here reflects my philosophy on how complex operations should be shaped into intelligent solutions and brought to life through the platforms built around them.

The focus is always the same: build strong foundations, reduce operational friction, and create systems that scale with clarity, resiliency, and intent.

Lightspeed Pipeline
01 // Data Orchestration

Self Hosted Prefect

Company-wide orchestration layer for ETL and AI workloads built on Prefect and EKS. Consolidated fragmented jobs into a single control plane, cutting infrastructure overhead by ~42% with serverless Fargate clusters, and eliminated concurrency-limit issues on worker nodes.

  • + Prefect
  • + AWS / EKS / Fargate
  • + Python
  • + Terraform / HELM Charts / GitHub Actions
02 // Platform Modernization

Perceptivity (DPaaS)

Architected, designed, and led development of the Perceptivity managed data platform — a fully cloud-native SaaS built and launched in under six months. Implemented end-to-end CI/CD, hosted core services on Google Cloud Run, and used Firebase + Firestore for auth, storage, and real-time application state.

Engineered a full TypeScript, Nuxt/Vue, and Node-powered stack with YAML-based workflow actions to support massive multi-tenancy.

Integrated an intelligent chat agent using Google Gemini (Genkit), enabling natural-language-driven insights, pipeline commands, and embedded AI application workflows.

The platform now serves 1,300+ clients and 3,000+ users, connecting to 590+ data sources and delivering 828 analytical models across our customer base. Perceptivity operates as the primary interface for data pipelines, insight delivery, and AI-augmented ops, forming the strategic core of Bear Cognition's managed data service.

  • + TypeScript / Vue / Nuxt
  • + GCP / Cloud Run / Firebase Auth / Firestore / Gemini Genkit
  • + Sigma / Tableau
  • + GitHub Actions
Perceptivity Platform
Bid Optimizer
03 // Optimization & AI

USPS Bid Optimizer

AI-enabled optimization engine for USPS contract routes.

Uses intelligent document processing to extract contract terms, then models client fleets (rates, overhead, margins, and risk tolerance) to calculate optimal vehicle allocation and pricing per route.

Built to maximize profit while increasing probability of award, with support for scenario analysis and what-if planning across large route networks.

  • + Intelligent Document Processing
  • + Python / Vue / Nuxt v4 / TypeScript
  • + Langflow / LangChain / OpenAI
  • + AWS / Lambda / Amazon Location Services
04 // OPEN SOURCE OPERATIONS

Open Source Engagements

Contributed to critical open-source projects supporting global engineering teams. Identified and resolved edge-case deployment failures, patched broken GCP logic, and improved CI/CD reliability across multiple repositories.

Submitted fixes to FirebaseExtended/action-hosting-deploy, enhancing deployment logging, addressing workflow failures, and improving hosting pipeline stability.

Located and resolved a defect in langflow-ai/langflow GCP deployment scripts, restoring Cloud Build functionality and improving installation reliability.

  • + TypeScript
  • + GitHub Actions
  • + Firebase Hosting
  • + Python
Open Source Contributions