Tamar Bercovici
HostVP of Engineering at Box
Ganesh Datta
CTO & Co-founder of Cortex
April 9, 2026
In This Episode
When Tamar Bercovici joined Box, the engineering team consisted of just 30 people. In the 15 years since, she's worked through every level of the leadership track in the organization and is currently the VP of Engineering. She leads the core platform, which is the backend layer that storage, search, metadata, and Box's AI capabilities all run on.
In this episode of Braintrust, Tamar and Cortex CTO Ganesh Datta get into what the job actually requires at each level from IC to VP, why the transition to a director-level role catches people off guard in a way the manager transition doesn't, and how platform teams make the case for their value when nothing they build has a user-facing feature attached to it.
You’ll learn
Tamar describes sitting down to write her first self-evaluation as a manager and genuinely not knowing what to put. What she did, or what the team did? That confusion resolves once you accept that what you're accountable for is the team's results instead of what you shipped.
As a manager, you're optimizing within constraints. As a director, you're responsible for the constraints themselves. If you see a structural problem and don't fix it, nobody else will.
You can't build a strategic roadmap without deeply understanding the architecture and trade-offs your team is navigating. That means staying close enough to understand what the team is actually hitting.
For infrastructure teams, the external end user is the primary customer. Internal engineering teams are secondary. Getting that hierarchy right determines how you measure success and how you make the case for headcount.
One engineer using AI effectively produces at the output level of several. The observability, deployment gates, and CI/CD practices that should have existed anyway become load-bearing under that pressure. Teams that already built those practices absorb what AI produces. Teams that didn't are about to find out.
Quotes
"I think what's interesting about AI coding is that it's really pressure testing what you already technically should have had in place."
Tamar Bercovici
VP of Engineering at Box
"Execution is burning down risk. That's a one-to-one. And so it's just going to manifest a little differently based on where in the stack you are."
Tamar Bercovici
VP of Engineering at Box
"If you give [AI] a bad prompt, you're going to get a bad outcome. Just like if a product manager gives you a bad set of requirements or your architect doesn't explain what they want you to do, the engineer's not probably going to do a good job."
Tamar Bercovici
VP of Engineering at Box
Timestamps
(01:36)
Tamar's role at Box and what "core platform" actually means.
(06:34)
How the job changes at each level (IC, manager, director, VP) and where each transition is actually hard.
(16:01)
Why Box has principal architects reporting at the same level as directors.
(18:35)
How to stay technically grounded as a senior leader without pretending you're still an engineer.
(24:29)
AI, reliability, and guardrails for a platform team where downtime has real consequences.
(29:45)
AI as an amplifier: why it makes strong teams stronger and exposes weaknesses in less mature ones.
(39:25)
How platform teams make the case for their value when nothing they build has a feature announcement attached to it.
Other episodes
Why DevOps Transformations Fail in Regulated Industries, with Merge Ready's Matt Bailey
Matt Bailey is a DevOps consultant and the founder of Merge Ready, a DevOps community and YouTube channel. He spends most of his time working with large regulated organizations across finance, healthcare, and government, helping them untangle the tooling decisions and processes that stall their software delivery.
In this episode of Braintrust, Matt and Cortex CTO Ganesh Datta dig into why buying a new CI/CD platform doesn't count as a DevOps transformation, what "decision latency" costs regulated organizations, and how to automate compliance.
May 21, 2026
Matt Bailey
Founder & Executive Producer at Merge Ready
Building a developer platform like a product: Inside The New York Times with Sneha Rao and Ahmed Bebars
At The New York Times, roughly 1,000 engineers build everything in the NYT ecosystem from breaking news notifications to games to the cooking app. A single team, ‘Developer Platforms,’ owns the infrastructure they all ship on, from CI/CD and runtime to SRE and FinOps. Sneha Rao leads the group as its VP of Product, and Ahmed Bebars is a principal engineer who works closely with Sneha.
In this episode of Braintrust, both Sneha and Ahmed, join Cortex CTO Ganesh Datta to dig into what it actually means to run a developer platform with a product mindset. They discuss how the team's rename from Delivery Engineering reshaped which conversations they were part of, and why centralizing both reliability and cost functions inside the Platform org leads to better decisions than keeping them separate.
May 7, 2026
Sneha Rao & Ahmed Bebars
VP of Product Management & Principal Engineer at The New York Times
Rootly's Dan Sadler on why AI coding tools are driving more incidents and why reliability is the product
In this episode of Braintrust, Cortex co-founder and CTO Ganesh Datta sits down with Dan Sadler, VP of Engineering at Rootly. Dan shares how building a business-critical product forced Rootly to prioritize operational maturity far earlier than most companies would, and why that discipline is now paying off as AI coding tools push incident volumes higher across the industry.
They also discuss what a genuine reliability culture looks like in practice, the specific cadences Rootly uses to stay ahead of production issues, and why Dan believes the rising pace of AI-generated code makes the infrastructure around code more important than ever.
April 23, 2026
Dan Sadler
VP of Engineering at Rootly
