pakkasys logo

Selected work

A few selected products and method notes from the collective.

Products

Selected in-house products we've been working on.

SmileyTap

One-tap feedback for physical spaces. Measure satisfaction where it happens.

For service desks, offices, and public-facing locations.

Role: product + backend + ops

Status: Live · Stack: Go + Postgres + Next.js

TilaFix

QR-based issue reporting for public spaces. Fast, anonymous, actionable.

For facilities, municipalities, and shared spaces.

Role: product + backend + ops

Status: Live · Stack: Go + Postgres + Next.js

RecSys

Auditable recommendation system suite with deterministic ranking and versioned ship/rollback.

For teams building personalized discovery, retrieval, and ranking workflows.

Role: architecture + backend + pipelines + evaluation

Status: Active · Stack: Go + HTTP API + offline pipelines + evaluation tooling

Open source

Selected repos that reflect our backend and reliability focus.

pureapi-core

Clean architecture foundation for Go services.

api-toolkit

Reusable Go building blocks for clean services.

api-svc

Reusable Go domain services for SaaS APIs.

logger (v2)

Async structured logging for production systems.

envvar

Strict config parsing for calm deployments.

randutil

Small randomness utilities for tests and tools.

Method notes

How we typically approach reliability work (not client-specific case studies).

Reducing alert noise, our approach

Focus
SaaS teams with on-call fatigue and noisy alerts.
Typical window
3-7 days
Signals
High alert volume, unclear thresholds, low signal-to-noise.
Approach
Review alert rules, align with SLOs, tune thresholds, add aggregation, trim duplicates.
Typical outputs
SLO draft, tuned alert rules, baseline dashboard, runbook skeleton.
Expected result
Calmer on-call, fewer false positives, clearer incident signals.

Stabilizing deploys, our approach

Focus
Legacy services with fragile pipelines and unclear ownership.
Typical window
1-2 weeks
Signals
Manual deploys, missing runbooks, inconsistent rollbacks.
Approach
Standardize the deploy pipeline, add runbooks, define rollback steps, baseline monitoring.
Typical outputs
Deploy checklist, rollback plan, pipeline tweaks, monitoring baseline.
Expected result
More predictable releases, lower regression risk, faster recovery.

If this matches what you need, most teams start with the Backend Health Check.

We can share details privately (NDA-friendly).