Nehemiah Onyango.
Software Engineer.
Product Builder.

PROBLEM

Updating my portfolio was surprisingly repetitive. Every new project meant editing markdown, swapping screenshots, committing, and deploying — the same five steps, every time, done by hand.

SOLUTION

A Python CLI that abstracts the whole workflow — content updates, branching, commits, optional pushes — into single, deterministic commands. Opinionated enough to stay consistent, extensible enough to grow.

RESULT

Updating my portfolio went from a fifteen-minute chore to a single command. Laid groundwork for CI/CD hooks and GitHub API integrations I haven't built yet.

LESSON

Developer experience is a product. If a tool only has one user, it still deserves the same care as one that has thousands.

PROBLEM

I had too many side projects living on disk — some active, some abandoned, some I'd genuinely forgotten existed. No single place told me the truth about any of them.

SOLUTION

A Flutter desktop app that indexes local projects automatically, using signature-based detection to identify frameworks and project types straight from the file structure, backed by a Drift/SQLite metadata store tracking lifecycle and Git health over time.

RESULT

A daily-briefing interface, closer to an intelligence dashboard than a file browser, that surfaces what's dormant, what's active, and what's quietly decaying — plus auto-generated portfolio artifacts from real project data instead of manual writing.

LESSON

Your filesystem already contains valuable product intelligence. Most developers just never build the thing that reads it back to them.

PROBLEM

Sports bettors track prediction slips in notes apps and spreadsheets, then never actually learn from their own history. There was no structured way to see which reasoning patterns were working and which ones were just noise.

ARCHITECTURE

A local-first data model in IndexedDB via Dexie, storing slips, prediction types, rationale, and whitelist/blacklist teams alongside historical outcomes — fully stateful, persisted client-side for speed and reliability, no backend round-trip required.

OUTCOME

A performance analytics view that surfaces which prediction patterns actually perform per league, plus structured note-taking that forces reasoning to be written down before the outcome is known — and export tools for sharing slips in social-friendly formats.

LESSON

People don't lack data, they lack a forcing function to record their reasoning before they know if they were right. That's a harder product problem than it looks.

PROBLEM

Goals die quietly. Not from one bad decision, but from thousands of small moments where nobody reminded you what you were actually working toward.

SOLUTION

A goal and dream reminder app that sends motivational push notifications and deliberately well-written notes, designed to keep people accountable to the version of themselves they set the goal for in the first place.

PROBLEM

Traditional contact apps are fundamentally broken. They store people as static entries — just names and phone numbers — organized alphabetically with no sense of time, context, or meaning. Users forget why they saved someone, networking value evaporates, and contact lists become unusable digital graveyards.

SOLUTION

Dossier reimagines contact management as a relationship intelligence system built on Intent Before Storage. Instead of saving and forgetting, users must pause and define why each person matters through a guided micro-reflection flow. Contacts organize by intent-based groups (Networking, Work, Friends, Opportunities) and display as Relationship Cards showing context, last interaction, and next suggested action.

ARCHITECTURE

Next.js PWA with offline-first localStorage via Zustand. Contacts store with metadata (reason saved, intent, tags, notes), interaction logs (type, timestamp, description), and insight layer (last interaction, follow-up suggestion, relationship score).

OUTCOME

80%+ of contacts categorized with clear intent within the first week. Active relationship management through follow-up prompts, network intelligence through streaks and achievements (Clarity Builder, Network Architect, Intent Master), and a dashboard that transforms passive lists into active relationship intelligence.

WHAT I BUILT

Turned a 175-response community survey into a ranked business-opportunity report with a 90-day validation plan, and a separate 102-respondent EXPO survey into a formatted strategic report — both used directly for decision-making, not filed away.

COMMUNITY PROFILER

Specified an internal tool to parse SaMaFa's 42-column Google Form dataset into structured community profiles — full PRD and parser spec, built to turn a messy form export into a reusable data pipeline.