waynele.
Let’s talk
← All work

TaloTrace / Autonomous QA

Software that tests software.

Agents that explore applications, run scenarios, and bring back evidence.

TaloTrace browser observation component showing a demo checkout, action cursor, and activity timeline

TaloTrace’s real LiveBrowserView component, captured in September 2026 and retained in the current webapp repository. The checkout is a fictional demo fixture. This shows the observation, cursor, timeline, and activity UI; it is not a customer session or a new full-app capture.

Try the public demo
My roleEngineering leadership & hands-on contribution
WhenApr 2026 – Present
FocusAutonomous QA
Stack
  • Python
  • Agno
  • FastAPI
  • Next.js
  • PostgreSQL
  • OpenTelemetry

Results & scope

Lower median browser-read latency
86.4%
Recorded experiment
223.3 → 30.5 ms in a regional placement probe on 6 October 2026.
Method & conditions

Three fresh browser allocations per region and 60 read commands per region. Excludes startup, model calls, and the full QA journey. Recorded experiment, not production-wide latency.

Lower median two-field execution time
72.8%
Recorded experiment
1.203 → 0.327 seconds with ordered calls in a controlled local experiment on 6 October 2026.
Method & conditions

Eight two-field groups per condition. Warm local browser and storage. Excludes model inference and production persistence. Not an end-to-end agent speedup.

Repositories with engineering guidance
14
Engineering scope
Shared engineering guidance integrated across fourteen repositories in the recorded standards rollout.
Method & conditions

Integration scope, not a count of products or teams. Hosted reviewer enforcement was not independently tested in that source review.

Authored Android benchmark scenarios
140
Engineering scope
Historical internal navigation benchmark spanning seven Android apps. The graded run scored 139 of the 140 authored scenarios.
Method & conditions

Scope is supported by integrated benchmark sources from May–June 2026. This is a coverage count, not a capability score; the internal assessment identified incomplete coverage and unmet release gates.

How I approached the work

Engineering decisions.

Measure the boundary before replacing the framework

Problem
Repeated remote browser requests amplified network round-trip cost inside an agent action.
Decision
Compared equivalent browser protocols, reduced redundant checks, and tested execution placement closer to the browser.
Result
The recorded placement probe reduced median read latency from 223.3 to 30.5 ms. I kept primitive latency separate from full-agent performance.

Group actions without losing recovery semantics

Problem
Independent field edits repeated observations and checkpoints. Grouping carelessly could replay uncertain mutations or bypass action budgets.
Decision
Used ordered action groups with pre-mutation budget reservation, durable outcomes, and explicit stop boundaries.
Result
The controlled local two-field benchmark fell from 1.203 to 0.327 seconds median, with field-state and checkpoint-recovery checks alongside timing.

Keep workflow ownership explicit

Problem
Agents, product interfaces, and platform services need consistent contracts and a clear owner for execution state.
Decision
Kept workflow state and budgets in the application while using native Agno components for agent execution. Introduced shared conventions and typed boundaries.
Result
Reusable execution foundations and engineering guidance were integrated across fourteen repositories. Native composition reached development with recorded smoke checks.

Inside the product

Following
the device.

The native observation component keeps the portrait device frame, current action, and activity timeline together.

TaloTrace native observation component displaying a fictional Talo Market app, touch marker, and activity timeline
Recorded component demo, September 2026. The device content is a deterministic fixture, not a customer session.

The work.

I help build TaloTrace, an autonomous QA product for web and mobile applications. My work connects agent navigation and execution with inspectable evidence and the customer interface. I now lead engineering delivery while staying hands-on with implementation and collaborative solution design.

My contribution

Delivery leadership

Turning requirements into engineering plans, reviewing important changes, and coordinating work through release.

Agents and evidence

Building across exploration, scenario execution, navigation, recordings, and reproducible issue traces.

Connected systems

Working across the product interface, backend orchestration, shared runtime contracts, observability, and deployment reliability.

Selected engineering contributions

What I built.
How I approached it.

37 documented contributions to TaloTrace. Start with these examples, or explore the full product index.

All 37 contributions
01
Built · integrated source

Booked crash-recovery wakes before committing work

Moved recovery scheduling ahead of transaction commit and shared workload-aware job waking and outbox senders across services.

Cloud Tasks · Pub/Sub · PostgreSQL · SQLAlchemy · Google identity

What I changed

  • Booked a job backstop before COMMIT instead of relying on an outbox sweep.
  • Shared outbox senders and the workload identity allowlist.
  • Handled the startup 403 returned by a newly created Cloud Tasks queue with a scoped retry.

The decision

A process crash after commit must leave a durable wake path; retry only the documented startup edge.

The result

Committed work has a booked recovery path without introducing a perpetual sweep.

Implementation and integration are verified. Current production status and customer outcomes were not independently checked.

02
Built · integrated source

Sealed test secrets in the browser and scoped identity access

Replaced the credentials UI with typed test-data items and user-answerable request cards while sealing secrets before browser requests leave the client.

Next.js · React · libsodium · NaCl sealed boxes · cryptography · PostgreSQL

What I changed

  • Added browser-side libsodium sealing before API submission.
  • Carried tenant identity with secret sealing and published the test-data public-key route.
  • Granted an identity window for each stored secret and prevented secret-like fields from leaking as ordinary editable properties.

The decision

Keep plaintext secret handling at the authorized boundary and make test requirements explicit items rather than hidden credential assumptions.

The result

Tests can reference saved logins/files/records, and the UI answers missing data in place.

Implementation and integration are verified. Current production status and customer outcomes were not independently checked.

03
Built · integrated source

Built recording replay with seek markers and tap overlays

Connected the run-review player to the real video playhead, action offsets and expiring artifact links.

Next.js · React · TypeScript · Vidstack · hls.js · TanStack Query

What I changed

  • Made seek an explicit player command and kept the timeline visible with step markers.
  • Overlaid the native tap point on replay.
  • Renewed an expired recording URL once while retaining playback position and synchronized the step panel to the playhead.

The decision

Use the actual playback clock and stored recording activity, not independent UI timers.

The result

Reviewers can move between recorded actions without losing their place when a signed link expires.

Implementation and integration are verified. Current production status and customer outcomes were not independently checked.

04
Built · integrated source

Imported documents as source-preserving passages and flow claims

Converted imported document passages into typed knowledge and flow claims while preserving source/version identity across edits and exploration imports.

MarkItDown · pgvector · PostgreSQL · Pydantic · Cloud Tasks

What I changed

  • Made one journey passage become one flow claim with its ordered steps.
  • Matched journey passages against flow claims instead of unrelated knowledge kinds.
  • Replaced a document’s own old words on update and minted exploration versions that keep their source.

The decision

Keep provenance and semantic claim identity stable when documents change instead of minting uncontrolled duplicate knowledge.

The result

Imported and observed product knowledge retain source ownership and flow semantics.

Implementation and integration are verified. Current production status and customer outcomes were not independently checked.

05
Built · integrated source

Made streamed assistant turns stoppable and replayable

Built durable live-event replay and made Stop cancel both an active provider stream and an in-flight tool call while retaining partial output.

Vercel AI SDK · SSE · PostgreSQL · SQLAlchemy · FastAPI · OpenRouter

What I changed

  • Supported idle stream close and durable live-event replay.
  • Used a bounded live-events table and coalesced deltas instead of treating the stream as the inbox.
  • Watched the turn cancel flag separately from lease renewal and persisted partial text with a stopped outcome.

The decision

Keep a durable transcript distinct from ephemeral delivery, and make cancellation interrupt the actual work.

The result

Users can stop a streaming turn without losing already received text or leaving an unanswered tool call.

Implementation and integration are verified. Current production status and customer outcomes were not independently checked.

06
Recorded development acceptance

Redesigned Android capacity around exact scale-to-zero leases

Moved from packed Android hosts to one device per VM and repaired capacity pressure during exact-generation retirement.

Redroid · Google Compute Engine · Pulumi · PostgreSQL · gRPC · agent-device · Appium

What I changed

  • Introduced one-device instances with no warm pool.
  • Reasserted capacity pressure after exact retirement so mixed demand does not oscillate.
  • Recorded development acceptance for ten simultaneous leases, replay-identical receipts, excess-capacity refusal and cleanup to zero.

The decision

Trade packed-host utilization for burst economics, isolated ownership and deterministic cleanup while keeping cold readiness visible.

The result

The saved development handoff reports correct ten-lease acceptance and final zero capacity.

Acceptance is a saved session report, not a rerun. Modeled cost savings and current production rollout are separate evidence.

07
Built · integrated source

Built iOS simulator lanes and native-provider readiness

Connected flexible iOS simulator execution with provider reconciliation and device-host provisioning.

Python · iOS Simulator · macOS · GitHub Actions

What I changed

  • Added flexible iOS simulator lanes with inventory, lease and host-control changes.
  • Implemented native-provider readiness and reconciliation as part of the mobile execution path.
  • Consolidated native lane ownership in the newer provider and removed the older mobile and Appium execution paths.
  • Changed host-provisioning checks to require native lane receipts after the earlier Safari qualification path was removed.

The decision

Make one native provider own the lane and gate host provisioning on evidence from that provider, keeping readiness checks aligned with the actual execution path.

The result

Integrated a progression from flexible simulator lanes to a consolidated native-provider path with matching provisioning checks.

The authored changes are integrated in source. The removed providers are historical design steps, not simultaneous current capabilities. No production device capacity, reliability rate or iOS store release is claimed.

08
Built · integrated source

Defined PostgreSQL ownership and transactional migration boundaries

Connected shared database primitives with service-owned migrations, schema checks and release phases.

Python · PostgreSQL · SQLAlchemy · Alembic · asyncpg · GitHub Actions

What I changed

  • Added shared PostgreSQL connection, transaction, readiness and repository primitives.
  • Ran Alembic changes, applied-ancestor tracking, grant reconciliation and catalog lint in one transaction protected by a PostgreSQL advisory lock.
  • Added service-owned migration adoption and grants, including identity and payload cleanup in the quality service.
  • Connected database owner configuration with release-phase guards and bounded migration execution.

The decision

Keep product schema ownership with each service while sharing the transaction and migration machinery that enforces readiness and catalog boundaries.

The result

Integrated shared database foundations and service migration adapters with explicit transaction and release-phase checks.

The authored changes are integrated in source. Repository ancestry and inspected implementation do not prove that every migration was executed successfully in production or that a release had no downtime.

09
Built · integrated source

Built sign-in, onboarding, billing and integration consent

Connected the customer entry points for TaloTrace with reusable screen states for design review.

TypeScript · React · Next.js · Firebase Auth · OAuth · MCP · Storybook

What I changed

  • Added Firebase sign-in, an authentication provider, route protection and bearer-token requests.
  • Implemented a multi-step onboarding wizard and shared onboarding context.
  • Added a credits and billing screen with a top-up gate before starting a run.
  • Built OAuth consent for MCP integrations, including safe redirect handling.
  • Added page-level Storybook stories and a screen-state export pipeline for design handoff.

The decision

Keep authentication and onboarding state shared, expose billing requirements at the run entry point, and make complete screen states reviewable in Storybook.

The result

Implemented the customer access and setup flow alongside billing, integration consent and reviewable product screens.

The authored changes are integrated in source. This describes the recorded implementation, not verified production adoption, conversion or a claim that every onboarding step was complete.

10
Built · integrated source

Made platform metrics and navigation readiness inspectable

Built staff dashboards that connect platform trends with evaluation scorecards and coverage.

TypeScript · React · Next.js · PostgreSQL · Storybook

What I changed

  • Added a platform-metrics screen with typed domain models, an API adapter, trend charts and screen stories.
  • Built a navigation release-readiness view with scorecards and coverage indicators.
  • Kept evaluation-store reads on the server in explicit read-only PostgreSQL transactions.
  • Rendered an unconfigured state when the evaluation data source was absent instead of substituting sample data.

The decision

Give staff a dedicated operating view and contain the evaluation-store exception behind server-side, read-only queries.

The result

Implemented reviewable metrics and readiness screens, including domain tests and explicit data-availability behavior.

The authored changes are integrated in source. This is operational visibility, not evidence of improved QA accuracy, system speed or a successful production release.

Explore the full TaloTrace contribution index
Next projectClevah