A live learning platform built and shipped in nineteen weeks
A funded startup needed a live-cohort learning platform before its first paid intake. We took it from scope to production in nineteen weeks.
19 weeks
Time to launch
From kickoff to the first paid cohort in production.
99.95%
Launch uptime
Across the first cohort's twelve-week run.
Under budget
Infrastructure cost at launch
Actual monthly run cost came in below the modelled estimate.
10x launch size
Concurrent learners tested
Validated by load testing before the first cohort started.
What was actually going wrong
- Industry
- EdTech
- Services
- Software Development, Cloud
- Technologies
- Next.jsDjangoPostgreSQLWebRTCAWSRedis
- 1
A hard launch date set by an already-marketed first cohort.
- 2
Live video, assessments, payments and progress tracking all needed for day one.
- 3
The founding team had no in-house engineers at the point of engagement.
- 4
Infrastructure had to stay affordable at low initial volume but scale for a tenfold intake.
What we did, and why in that order
Sequence mattered here as much as the individual decisions. Each step below made the next one safe.
- 01
Cut scope to what the first cohort truly needed
A discovery workshop separated launch-critical features from those that could wait, which is what made the date achievable.
- 02
Buy before building
Payments, email and video infrastructure were integrated rather than built, reserving engineering effort for the learning experience itself.
- 03
Ship weekly to a live staging environment
The founders used the product every week from week four, so course corrections happened in days rather than at a final UAT.
- 04
Design infrastructure for both ends of the range
Serverless and managed services at low volume, with a documented scaling path validated by load testing at ten times the launch cohort.
Platform architecture at launch
- Next.js frontend on CloudFront
- Django REST API on ECS Fargate
- PostgreSQL RDS with read replica
- Redis for sessions and caching
- WebRTC live classroom service
- S3 and MediaConvert for recordings
What changed, measured
Each figure below was measured after the engagement, not projected during it.
19 weeks
from kickoff to first paid cohort
Time to launch
From kickoff to the first paid cohort in production.
19 weeks
Launch uptime
Across the first cohort's twelve-week run.
99.95%
Infrastructure cost at launch
Actual monthly run cost came in below the modelled estimate.
Under budget
Concurrent learners tested
Validated by load testing before the first cohort started.
10x launch size
PLACEHOLDER: client quote pending written approval. They cut the scope we did not need and hit the date we could not move.
The capabilities behind this engagement
Facing something similar?
We will tell you what we would repeat from this engagement, what we would change, and what it is likely to take.