Projects

Engineering work across cloud platforms, reliability, and production AI - with clear labels for portfolio demos versus case studies.

Deployed project

Askel Job-Search Workspace

Job seekers lose track of applications and interviews buried in their inbox. I built and run Askel, a production app that reads Gmail and Calendar, proposes board updates for review, and keeps every application, interview and contact in one place.

Live at getaskel.com with separate staging and production, 495 merged pull requests since Aug 2026
3,467 automated tests, plus an email classifier checked against a set of 7,000 real recruiting emails

Next.jsTypeScriptReactAuth.jsNeon Postgres

Read case study Demo

Deployed project

InboxLink Gmail Connector

Every app that reads Gmail rebuilds the same OAuth, token refresh and message parsing. I built InboxLink, a standalone connector service with a typed SDK, so an app gets mail through one small API.

Open source, with @inboxlink/core and @inboxlink/sdk published on npm
Powers Askel's mail reading in production; 92 merged pull requests since Sep 2026

TypeScriptNode.jsNext.jsGmail APIOAuth 2.0

Read case study Code Demo

Public engineering portfolio

efokou-labs GitOps Platform

Scattered demos lived in one-off repos and cloud accounts. I built a reproducible lab with kind, Argo CD, Terraform modules, and sample workloads so any engineer can bootstrap a full stack from a fresh clone.

One kind cluster deploys Next.js, FastAPI, Postgres, and Redis through Argo CD overlays
Terraform modules run create-and-destroy tests in CI; AWS demos are torn down after each run

KubernetesArgo CDTerraformAWSGitHub Actions

Read case study Code Demo

Technical case study

Enterprise Observability Rollout

Production microservices had fragmented metrics and logs, slowing incident response. I led rollout of Prometheus, Grafana, ELK, and PagerDuty alerting across distributed services.

Unified dashboards and alerting improved detection and resolution across service boundaries
Centralized log pipelines in ELK and CloudWatch shortened root-cause analysis during outages

PrometheusGrafanaDatadogELK StackAWS CloudWatch

Read case study

Technical case study

Cloud Infrastructure Automation

Manual provisioning and inconsistent releases blocked multiple engineering teams. I standardized Terraform, Ansible, and CI/CD pipelines for auditable, repeatable delivery.

IaC replaced manual environment setup with version-controlled, repeatable provisioning
Shared Jenkins and GitHub Actions patterns increased release cadence across teams

TerraformAnsibleAWS ECSDockerJenkins

Read case study

Reference architecture

Agentic AI Platform

LLM workflows needed grounding, guardrails, and ops visibility. I built orchestration for tool use, RAG, and telemetry wired into existing Prometheus and Grafana stacks.

Multi-step agents with tool routing and human-in-the-loop checkpoints for reliable completions
RAG and evaluation harnesses grounded answers in internal docs; latency and cost tracked in production

PythonLangChainOpenAI APIRAGVector DB

Read case study Code

Public engineering portfolio

Agentic AI Evaluation Benchmarks

Coding agents need tests that can't be gamed. For Snorkel AI projects I built coding benchmarks with containerized environments, reference solutions and hidden tests, and evaluated OpenAI and Anthropic coding agents on them.

Containerized environments, reference solutions, hidden tests and reproducible verification for each task
Agent runs analysed with Harbor/Terminus: execution traces, test outcomes, token usage and cost

PythonDockerpytestHarborTerminus

Read case study Code