Philipp Gross Kochnov
Engineering & Delivery Leader
7 years in tech · 5 years leading teams · Flexible: remote, relocation, or on-site for the right thing.
I walk into delivery systems that are on fire and hand them back shipping predictably, no heroics required. Same discipline for ten years, whether the executor is a human team or AI agents. Pick the lens you're hiring for; the engine underneath doesn't change.
Same pattern, every time.
NDA · Crypto
2024 – Present · 1.5 yrs
Engineering Manager · Quality & Delivery · Singapore · Remote
Owned delivery and release readiness for a regulated 24/7 crypto product with direct financial impact.
Ran an AI-in-development pilot in production: one managed engineer delivering multiples of a conventional engineer's output, quality metrics flat.
The single escalation point for delivery risk and cross-team dependencies on a regulated, real-time product.
Cut production incidents by more than half and turned release readiness from a gamble into a repeatable process.
- Routine production incidents cut by more than half
- Release acceptance from ~1 week to 2 days
- Every bug driven to root cause, then a fix that holds
- AI-in-dev pilot: one managed engineer, output multiplied, quality flat
The full story
Five QA leads before me had failed to steady the product or the team. The environment ran on fear: fires in production, missed dates, features arriving as something nobody ordered.
I became the team's shield. That meant a controlled conflict with product and project management, fought in the one language they could not argue with: risk and data. Not "no," but "here is what happens to capacity, velocity and switching cost if we pivot now." Through iterations I earned carte blanche over the processes of a 30-person team.
Then we rebuilt it: git flow and feature flow, postmortems and bug-mits on every defect, shift-left across the pipeline. Critical incidents went to near zero. And I ran a live experiment putting AI into the development loop, where one managed engineer delivered roughly triple the output with the quality bar unchanged.
Stenn Technologies
2022 – 2024 · 3 yrs
Engineering Manager · Platform & Quality · London · Remote
Owned delivery speed across the product org and moved teams off a central release gate onto shipping independently.
Built the company-wide automation and tooling that took release regression from days to about 20 minutes.
Drove a shared delivery model across nine product teams while the company scaled from 5 teams to 9 and 90 people to 140.
Rebuilt quality from a manual release bottleneck into an automation-first system teams actually shipped through.
- Release cycle from 4-7 days to 20 minutes
- Org grew 5 to 9 teams, 90 to 140 people, delivery accelerated not slowed
- Directly managed 14 engineers and 2 leads, promoted ICs into leads
- Quality Guild Lead across 9 teams: automation-first, CI/CD
The full story
Stenn was scaling fast and the central regression queue was the architectural bottleneck. You could not just speed it up, you had to change the model. And selling that head-on would not work.
So I found two teams willing to pilot, rolled out automation under my team's watch, got hard numbers, and used those numbers to sell the change to everyone else. I wrote the test pyramid not as a QA document but as an agreement across all of engineering about who owns quality at which layer, and built custom monitoring: daily quality and delivery metrics per team, a weekly digest. One look at the dashboard and a team knew what was off.
Zonatelecom
2020 – 2022 · 2 yrs
Quality Team Lead · Russia · Remote
Turned a fragmented practice into a repeatable delivery cycle and became the coordination point across engineering, product and support.
Wrote a custom end-to-end automation framework for IP telephony (SIP, RTP, Asterisk) with no equivalent online in 2021.
Owned release coordination across UI, API and VOIP domains and made shipping predictable for the first time.
Built test automation from zero and cut critical production incidents by 3.4x.
- Critical production incidents down 3.4x
- Built a custom end-to-end framework for IP telephony (SIP, RTP, Asterisk), none existed in 2021
- Grew the team 3 to 5, automated mobile and the portal
The full story
A system running calls, video and mail, with zero tests and requirements that arrived in whatever shape. Every release shipped critical incidents, and the IP telephony went down for hours at a time.
I built the e2e automation framework for telephony from scratch, grew the team, automated mobile and the portal, and put structure where there had been none. Critical incidents dropped 3.4x.
Before management: engineering roles in test automation and quality (2018 – 2020).
Zonatelecom · QA Automation Engineer Feb – Nov 2020
Python / Pytest automation for UI and API, load testing with Locust, and the reusable test utilities and docs the team relied on after I moved into the lead role.
eBlitz · QA Engineer (Contract) Aug 2020 – Mar 2021 · London
Manual and automated testing of mobile apps (iOS, Android). Regression and exploratory testing focused on stability and usability.
Daedu · QA Engineer May 2018 – Aug 2020
Where it started. UI, API and mobile testing for web products. Built the early automation skills and learned how quality fits into a delivery cycle.
One engineer with AI, three times the output
1 engineer + AI 3× output
One engineer's output, autotest coverage plus a frontend e2e test framework, tripled in the same time period once AI ran the autonomous middle. Quality metrics flat, in a live production team, not a benchmark: AI managed as a delivery system, not used as a toy.
Then I tested the thesis on myself.
I designed an AI delivery pipeline and shipped real products end to end, architecture to app stores to first paying users, without writing a line of code by hand. That result was AI run as a managed system. So I ran the whole pipeline solo: real products shipped end to end, architecture to app stores to first paying users, no code written by hand. I ran the entire delivery system solo: real products from architecture to app stores to first users, one person owning scope, risk and dates while AI did the execution. I tested it on myself, quality bar included: real products shipped end to end through an AI pipeline, with the gates and release discipline that keep them stable. Two domains on purpose, to prove the method adapts and isn't a one-off.
-
Hiker's Voice The first product I built end to end on AI with no code written by hand. Proof the approach works at all. Case → Live ↗ -
Dreambook Where I built the shared platform and the content factory. A one-off app became a reusable system. Case → Live ↗ -
Dating Coach The platform run through a harder domain, plus all the content, sites and traffic. The engineering repeats; the domain and its go-to-market are the real work. Case → Live ↗ -
Quarterdeck The tool I built to run my own agent fleet. It watches every Claude Code session and lets the agents reach back. Case → Live ↗
Facts: availability, languages, flexibility
03 · Facts- Availability
- Full-time · fractional · advisory · consulting
- Languages
- Russian (native) · English (working / professional)
- Flexibility
- Remote · open to relocation · on-site for the right thing
What people who reported to and managed me say.
Philipp has a unique ability to identify and solve issues in software development processes, significantly improving how our products ship. His leadership and constant drive to raise the bar inspire other team members, and he unites every stakeholder around a shared standard of delivery.
Under Philipp’s guidance our team significantly improved our automation and delivery. He motivates the team and fosters professional growth, and his communication ensures effective interaction with both technical specialists and business stakeholders.
Phil led his team through complex projects with precision. He identifies issues early and implements process improvements that streamline workflows, reducing defects and time-to-market, all while being an excellent mentor and motivator.
All recommendations on linkedin.com/in/philya
A modern delivery system runs itself. You set the frame and own the gates, AI runs the autonomous middle, and every return fixes the input. Delivery that compounds.
I'm looking for the next broken delivery system to rebuild, this time AI-native. I'm looking for the next team that wants AI run as a delivery system, not a toy. I'm looking for the next program that needs delivery made legible end to end. I'm looking for the next quality function to turn into a delivery accelerator. As your next leadership hire, or fractional, advisory, or a consulting engagement. Flexible on location: remote, relocation, or on-site for the right thing.