Oddform Technologies

Problem
Oddform Technologies had a product idea so abstract their pitch deck could double as a Dadaist poem. Internally, they were juggling data from three SaaS platforms, a custom lab app, and a founder’s “vision board” made in Notion. Nothing talked to anything else — which meant staff spent more time copy-pasting than building the future.

Intervention
SP/CTRL took the chaos and wired it into a coherent, privacy-conscious workflow. Relay mapped the core data flows and designed a webhook bridge between the SaaS apps and the lab’s custom tool. Cronica built a cron-based job runner to keep everything in sync without human babysitting. Delta spun up a prototype bot to handle edge cases — logging anomalies and poking the right human when the system couldn’t decide what to do.

Some parts are already live (see the demo below), while others are still in the works. The backend gremlins are operational, but still learning their manners.

Glyphs & Glitches
API bridge that transforms inbound data before it hits the lab tool, so formats are always right. Cron job scheduler with self-healing retry logic (no more “we missed the 2 AM run” drama). Private logging system running locally — no customer data leaves the lab. Prototype Slack bot that escalates only when rules fail, cutting noise by 80%.

Tools
Python + FastAPI for the webhook handler
PostgreSQL for internal queueing
Linux cron for scheduling
Custom logging script with rotating log files
Slack API for targeted alerts

Impact
Data silos collapsed into a single flow, saving the team ~12 hours/week in manual work. System reliability jumped from “we think it ran” to a verified 99.8% success rate. Founder morale: restored. Maybe.

Who Helped
Relay, Cronica, Delta.

Oddform Technologies is a work in progress. We built a decoupled automation loop: the page enqueues tasks, a cron-driven worker processes them, and the UI reads results from a feed. Health checks and logs keep it observable; nginx + systemd (in prod) keep it reliable. It’s a small demo that mirrors a real job queue + worker architecture.

No AI magic — just clean, working DevOps you can build yourself after a weekend and enough coffee.

• Queue boundary: POST creates durable work; no long HTTP waits.
• Worker cadence: cron + processor isolates spikes.
• Observability: /health, logs, deterministic feed.
• Operability (prod): nginx routing, systemd supervision, port 8000 closed.
• Safety: static frontend; backend isolated behind proxy.