Bliro

A founding-stage company has more workstreams than it has people to run them. Someone has to hold the scope, the sequence and the date for each one, and keep them from colliding.

Took project lead on several tracks at once and ran each from scoping to delivery: billing and entitlement management, usage tracking, compliant voice recognition, the internal management system, and the enterprise-readiness work that gets Bliro through procurement. For each one that meant setting scope with the founders, cutting it into deliverables, sequencing it against everything else in flight, pulling in the departments who live with the result, and staying accountable for the date. Most of that job is not code. It is deciding what gets built when, and saying no to the rest.

A founding-stage desktop app has to be dependable before it can be clever. That means a real audio engine, a transcription pipeline that survives bad networks, an architecture the team can keep building on, and a way to turn production signal into the right work in the right order.

Took platform lead and rebuilt the desktop app from the building blocks up: reworked the audio engine, built a resilient transcription socket that holds through flaky connections, set the architecture, and wrote the conventions the team now builds against. Then closed the loop with an agentic workflow that pulls from Sentry, Amplitude, and other metrics, shapes it into candidate work, and feeds the backlog on a regular, reviewable cycle. User feedback runs through the same pipe: triaged into planned releases with the board and the delivery manager, so what ships is what actually moves stability and performance. Owns the whole release-management workflow too: separate tracks, fully automated with GitHub Actions and code signing, so a release ships from a single commit scoped to patches or minors, or a one-off workflow run.

A small team loses what it knows when people move on, and new joiners take too long to get productive without a shared foundation.

Started the weekly knowledge-sharing sessions and set the good practices, conventions, and documentation that give new joiners a foundation and make it easier to change how the team works later.

A small team loses what it knows when people move on, and new joiners take too long to get productive without a shared foundation.

Started the weekly knowledge-sharing sessions and set the good practices, conventions, and documentation that give new joiners a foundation and make it easier to change how the team works later.

A product that charges money needs billing that does not flinch, plus the unglamorous backend building blocks everything else depends on without noticing.

Engineered Bliro's billing system and several of the core backend building blocks underneath the product: the plumbing the rest of the app relies on every day.

The team's daily work was full of repetitive, error-prone steps, and managing the data by hand was slow and easy to get wrong.

Built an internal management system that lets employees manage the data and finish the repetitive parts of their daily job significantly faster and more deterministically. Shaped it by pairing with the departments that live in it, customer success, the delivery manager, and other platform owners, so it carries the features each of them actually needs.

Maintenance is the work that never trends. Crash fixes, flaky edges, the slow decay a shipping desktop app collects. It is essential, easy to defer, and deferring it is exactly how a product turns fragile.

We built an agentic maintenance workflow: the agent does the work, the team supervises. It runs on a small set of commands we drive, /goal to point it at what matters and /loop to keep it iterating, wired through agentic workflows that read production signal, choose the work, and open reviewable changes. /goal sets the target for the week, from Sentry and Amplitude signal to a specific reliability or performance goal; /loop keeps the agent working the problem, proposing changes, running the checks, and handing back only what passes. Every week it delivers a batch of finished work that we review, merge, and ship. Nothing merges without a human reading it, so the agent moves fast and we stay accountable. Performance climbed, the app got steadier, and customer feedback stopped piling up, because it gets triaged and closed inside the same loop.