engineering manager · london
Gunjan Kumar Mehta
I lead engineering teams that ship — from hands-on architecture to leading through managers, across countries and time zones. 9+ years engineering, 4.5 leading.
currently: building ai-driven insurtech at concirrus
about
Leadership, hands-on
Good engineering leadership is a balance of speed, quality, and tech debt — making the calls that let teams move fast without losing trust in the system. I lead from inside the work: driving architecture, reviewing code, and owning delivery end to end.
I've been curious about computers since I was a kid — and here I am, a CS engineer. Nine years on, that curiosity has taken me from building Java microservices in Jaipur to leading engineering managers in London, with teams spread across five-plus countries and time zones.
These days I spend most of my energy on one question: what does a great engineering team look like when AI writes a lot of the code? The answer we keep landing on — humans who specify, review, and verify.
outside work: swimming · teaching · exploring new places
Lead by example
I still design, code, and review. Credibility comes from the work.
Specify, review, verify
AI writes more of the code now; humans own the judgement.
Teams are systems
SOLID principles apply to squad design as much as class design.
Alignment before architecture
Event Storming with stakeholders before a line of code is written.
Grow people deliberately
Structured development plans — 15+ engineers mentored to promotion.
9+
Years in software engineering
25+
Engineers hired across teams
15+
Engineers mentored to promotion
3x
Delivery velocity with AI-assisted engineering
featured work
Case studies
Three stories of teams, systems, and the results we shipped together.
case study 01
AI submissions extraction at scale
Manual submission processing was a bottleneck across four lines of business. We built an email-ingestion pipeline with OCR and document AI that turned unstructured broker submissions into structured underwriting data.
case study 02
Rolling out AI-assisted engineering
We moved every squad from "write" to "specify, review, verify" — Claude Code and Cursor for coding, CodeRabbit for review — with team norms for review standards, attribution, and security of AI-generated code.
case study 03
Scaling the team, not just the system
From directly leading 10–15 engineers to leading through 3–4 Associate EMs, each owning a domain squad in a conveyor-belt delivery model — shipping features in parallel without losing quality.
latest writing
Notes on engineering leadership
From "write" to "specify, review, verify"
How we re-skilled every engineer for AI-assisted delivery — and what changed about hiring, junior roles, and performance reviews.
Applying SOLID principles to team design
Single responsibility isn't just for classes. What happened when we restructured squads the way we structure code.
Event Storming to 100% alignment
5–10 new ideas per session, natural service boundaries, hidden complexity surfaced early.
let's connect
Open to interesting problems
Recruiting conversations, speaking, mentoring, or consulting on the technologies and teams above — I'd love to hear from you.
gunjanmehta61@gmail.com