How a Startup is Changing a Legacy Org from the Inside

Three and a half years ago, our 17-person startup was acquired by a 4000+ person organisation with a 30-year-old technology estate. We were placed in a new city, a new hub, and an org where teams wouldn't even give us read access to their codebases.

This is the honest story of how we went from outsiders to partners — and how a small team with zero positional authority shifted engineering culture across an organisation that had accumulated decades of process debt.

What we got right: Starting with empathy instead of expertise. Understanding that past decisions were rational given past constraints. Earning trust through small wins before suggesting any change. Starting with the willing — not trying to boil the ocean.

What we got wrong:
Trying to introduce an Internal Developer Portal before we had the political capital to navigate process debt — we couldn't even get a VM. The frustration and near-burnout when the org wouldn't let us contribute.

The turning point:
A RAG system built across 8 teams — 16 languages, 25 countries — that became the first true cross-team collaboration in the org's history. No Agile lead, no scrum ceremonies. People just solved a real problem together.

The numbers:
Deployment frequency from once every 2 months to every 10 days. Build times down 80%. Excel-sheet environment booking replaced with self-service ephemeral environments. Every new service ships with monitoring, observability, docs, and tests from day zero.

The culture shifts:
We flattened PR reviews so every dev — not just the EM — gives feedback. Devs and QAs now present directly to stakeholders across 25+ squads. Decision-making moved from a Tech Lead bottleneck to distributed RFCs where anyone can propose and comment. We removed executive offices — everyone sits on the open floor. We introduced 360-degree reviews where none existed. And we built a 2000+ person tech community from zero.

Where we are now:
Experimenting with 3-person pods working in flow beyond sprints, leveraging AI, solving customer problems daily. Invite over inflict — the teams that adopted these ways of working did so because they saw the outcomes, not because anyone mandated it.

This talk is grounded in systems thinking, game theory, Team Topologies, InnerSource.

Mostly the talks is around the learnings of hardest engineering problems aren't technical — they're human. AI can write the RFC. It can't build the trust that gets it accepted

Takeways: Will leave with practical patterns for influencing organisational change without authority — whether they're a small team inside a large org, or a single leader trying to shift how their corner of the world works.

Jeevan D C
Solving tech problems at scale
A problem solver keen on Distributed Systems, System Architecture, Scale, Performance with core expertise in building high throughput, resilient, fault-tolerant applications in multi-cloud environments.
Passionate about teaching and mentoring. Got extensive experience in Digital transformation, Cloud migrations, Startup acceleration in Telco, Insurance, Banking domains at scale.