About Blog Posts Projects Contact LinkedIn

My name is Chandra Vijay.

I build products. All the way down.

Chandra Singh

Twelve years ago my job was to break software before users did. I got pretty good at it.

What I didn't expect was how much that thinking would follow me through every role after. I moved from QA into engineering leadership, then into product management — and ended up doing both seriously, not just dabbling. The job description kept changing. How I approached problems didn't.

These days I lead quality engineering and own consumer products used by 10 million people across Europe. I write here about what that work actually looks like from the inside.

Quality Engineering
Hiring for QA or SDET?
Lead SDET QA Manager QA Tech Lead
See my QA work →
Product Management
Hiring for Product?
Platform PM Technical PM AI PM Product Builder
See my PM work →

Blog Posts

Words are
being sharpened.

Twelve years of product thinking, war stories, and hard-won frameworks — currently being turned into writing worth your time. Check back soon.

● Building

Twelve years. Two crafts.

↓ QA Resume ↓ PM Resume

Twelve years ago I was writing test cases for a hotel management system nobody loved. Today I lead quality engineering and own consumer products for OneApp — Deutsche Telekom's platform running across nine European countries for 10 million monthly users.

For a long time, the most interesting part of QA work wasn't the bugs I found — it was the decisions behind them. Bad flows. Assumptions nobody had questioned. Features built for the wrong user entirely. I kept asking why those decisions got made, and that question pulled me gradually further up the stack, into engineering leadership and eventually into product.

Coming from engineering changed how I do product work. I've seen too many specs fall apart between planning and build to write requirements that leave room for guesswork. Discovery isn't optional when you've watched, firsthand, how much a feature can drift from its intent between the PRD and the demo. It's not a different perspective — it's just more of the picture.

The thread through all of it: I care more about whether something works for the person using it than whether it technically shipped.

12
Years across quality engineering and product management
10M+
Monthly active users on products I've shipped and managed
9
EU markets · Germany, Greece, Croatia, Hungary and five more
5
Companies across healthcare, OTT, payments, and telco

I started in test automation when Selenium grids had to be configured by hand and documentation for most tools was either thin or nonexistent. The problems were different then — more manual, more fragile. Now I lead quality engineering for a product serving 10 million users across nine countries, with a team of 10+ engineers. The challenges are different too. Figuring out what "good" looks like when the output is generated by an LLM isn't something most QA teams had to think about until recently.

At Deutsche Telekom, OneApp runs on iOS and Android across nine EU markets at the same time. Different regulatory requirements, different release cycles. When I joined, automation coverage was 40%. I built it to 72% and integrated it directly into CI/CD — quality gates on every build, not a review at the end of the sprint.

We also had a performance problem. Crash-impacting sessions at 2.8%, cold start at 4.2 seconds, app store rating sitting at 3.6. I ran a focused program on both metrics: crashes down to 0.9%, cold start to 1.8s, rating up to 4.2. Small numbers on paper; noticeable on a platform used daily by millions.

When LLM features started shipping — AI assistant, search, engagement — I had to figure out how to test something without a deterministic right answer. I built an evaluation framework using golden datasets and LLM-as-judge scoring for response quality, faithfulness, and safety. It runs in CI/CD. It's not a quarterly audit; it's a gate on every deployment.

Before Deutsche Telekom, I spent four years at 3Pillar Global on ModMed — US healthcare SaaS, regulated, high-stakes. Last 18 months: zero P0 production defects. That came from building the right processes, not from being careful.

If you're building a product where quality actually matters — let's talk.

Tools
Selenium Appium Playwright Cypress RestAssured Java Python Jenkins CI/CD LLM Evaluation RAG Testing LLM-as-Judge AI Safety Guardrails Grafana Kibana JIRA
Domains
B2C Mobile Healthcare SaaS OTT Payments EU Regulated Markets AI/LLM Features
72%
Automation coverage at Deutsche Telekom — up from 40%. Industry average sits at 30–35%.
4.2★
App store rating after performance program — was 3.6
−68%
Crash-impacting sessions, from 2.8% down to 0.9%
0
P0 production defects over 18 consecutive months — regulated US healthcare platform

I came to product through eight years in engineering. I've written the test plans, led the automation teams, sat in sprint reviews when features landed wrong — and knew exactly why. By the time I wrote my first PRD, I already had a clear picture of where product decisions actually break down. Not in the roadmap. In the platform underneath it.

I moved into product at Deutsche Telekom in 2022 and today own the platforms and systems that power OneApp — the release infrastructure, notification stack, auth layer, and AI backbone that 10M+ users touch daily across nine European countries. Nine markets, one platform, constant change.

My first PM project was fixing the release process — it was the bottleneck for every engineering team in the org. I root-caused it through stakeholder interviews and release data, then built a Release Management Platform that standardised workflows across all teams. Go-to-market time: 4 weeks to 3. Patch releases: down 35–40%. Documentation overhead: cut 40% via automation.

From there I owned the Notification & Engagement Platform. Four fragmented systems across eight EU markets — everyone assumed the problem was content and targeting. It wasn't. The platform itself was broken. I reframed the brief, built a 3-quarter consolidation roadmap, and led cross-functional delivery across product, engineering, design, CRM, and eight national teams. Four systems became one. Integration overhead dropped 60%. DAU/MAU moved from 18% to 27%.

I also owned Auth & Onboarding end-to-end. Eight markets each asked for their own login flow. I said no — unified UX compounds long-term. Got product, security, legal, and operations aligned across all markets. Shipped OTP, auto-login, and entry-point simplifications that drove a 9% conversion uplift on login and kept us fully GDPR compliant.

Most recently: Ask-T, Deutsche Telekom's AI assistant for conversational support. AI PM is different — you can't write "it works" in the acceptance criteria when the output is generated. I defined user journeys and what good looks like for answer relevance, fallbacks, and escalation paths. Then built the evaluation layer: sample queries, expected patterns, edge cases, feedback loops. Twelve years of defining "done" in QA was the right preparation for this. The question is the same. The answer just moves.

Looking for a Platform PM, Technical PM, or AI PM? Let's talk.

Skills
Platform Strategy Technical Specifications System Design AI Product LLM Evaluation Developer Experience API Integration Stakeholder Management Cross-functional Delivery Executive Communication Go-to-Market OKRs PRDs Roadmaps User Research Behavioral Analytics SQL Figma Mixpanel Amplitude Jira Notion
Domains
Consumer Apps Platform Products Engagement & Notifications Auth & Onboarding AI/Conversational UX EU Multi-market 0→1 Builds B2C Telco
4→1
Notification systems consolidated — four fragmented platforms into one unified infrastructure
−25%
Go-to-market time after Release Management Platform — 4 weeks to 3
27%
DAU/MAU after platform consolidation — up from 18%
−40%
Fewer patch releases after release infrastructure rebuilt

Experience

Deutsche Telekom Digital Labs
Mar 2022 – Present
Gurgaon, India
Product Manager — Home & Core Platform 2023 – Present
Notification & Engagement Platform

Four fragmented notification systems. Eight markets. Zero shared infrastructure. I didn't start with the solution — I started by questioning the problem. Everyone assumed low engagement was a creative issue. After synthesizing behavioral data and interviewing stakeholders across 8 NatCos, the real problem was the platform itself. So I reframed it: stop optimizing, start consolidating. Built the 2025 platform strategy and 3-quarter roadmap. Led delivery across product, engineering, design, CRM, and 8 national teams. One unified framework replacing four legacy systems.

  • Push reach: 1.39M → 2.1M users (+50%)
  • Push CTR: 5% → 9% via ML-driven personalization
  • DAU/MAU: 18% → 27%
  • Integration overhead: down 60%
Authentication & Onboarding

Eight markets, each asking for their own login flow. I said no. Not because local needs don't matter — they do. But unified UX compounds. I ran the numbers, made the call, and got product, security, legal, and operations aligned across all NatCos. Shipped OTP, auto-login, and entry-point simplifications that worked everywhere while staying fully GDPR/CCPA compliant.

  • Login conversion: +9%
  • Onboarding funnel penetration: +5%
Ask-T — AI Assistant (Conversational UX)

Helped define what good looks like when the output is generated, not programmed. Worked on product discovery and experience definition for Deutsche Telekom's AI assistant for conversational support. Translated user needs into requirements, user journeys, and acceptance criteria. Defined how the assistant should handle answer relevance, fallbacks, escalation, and user trust. Built the evaluation layer — sample queries, expected answer patterns, edge cases, qualitative feedback loops.

  • Metrics defined: query completion rate
  • Fallback rate, escalation rate, CSAT/NPS
  • Repeat usage and support deflection
Insights & Experimentation Infrastructure

Built an AI-assisted workflow that turned marketplace reviews, behavioral data, and qualitative feedback into prioritized opportunity briefs — used to validate or kill roadmap candidates before engineering touched them. Designed experimentation workflows that let product and growth teams move faster with less engineering dependency.

Product Manager — Platform Engineering Mar 2022 – 2023
Release Management Platform

The release process was the bottleneck. Four weeks to ship anything. I root-caused it through stakeholder interviews and release data, then built a platform that standardized workflows across teams.

  • Go-to-market: 4 weeks → 3 weeks (25% faster)
  • Patch releases: down 35–40%
  • Documentation overhead: down ~40% via Notion AI
Technical Lead — Quality Engineering Jul 2022 – 2023
  • Manage and mentor 10+ SDET/SDE engineers across multiple pods — set quality standards, run code reviews, drive engineering excellence
  • Led QA for AI-powered features (LLM Chatbot, AI Search, AI Engagement Engine) — built golden-dataset evaluation framework using LLM-as-judge to validate response quality, faithfulness, and safety
  • Scaled E2E automation coverage from 40% to 72% across iOS and Android, integrated into Jenkins CI/CD
  • App performance program — crash-impacting sessions 2.8% → 0.9% (68% reduction); cold start 4.2s → 1.8s; store rating 3.6 → 4.2
  • Built AI-Powered Monkey Testing tool to proactively detect app crashes — improved stability by 20%
  • Led Germany NatCo go-live and COSMOTE app launch — owning final quality gates and cross-functional sign-off
  • Built internal AI agents using Claude and GitHub Copilot to automate QA documentation and release reporting — reduced overhead by 40%
3Pillar Global
Mar 2018 – Mar 2022
Noida, India
Module Lead — Quality Engineering

ModMed is a regulated US healthcare SaaS — the kind of product where quality failures have real consequences. I spent four years there, built out the QA function, and by the end we were shipping quarterly releases with zero P0 production defects. That didn't happen by accident.

  • Led QA strategy for ModMed — regulated US healthcare SaaS platform handling 3M+ patient interactions
  • Managed team of 5 engineers; drove quarterly releases with zero P0 production defects in final 18 months
  • Increased automation coverage from 35% to 68%; built RestAssured + Java API automation framework
  • Led quality for Electronic Medical Assistant (EMA) module — UAT sessions validating product against regulatory requirements
  • Served as SME for MMPM module — partnering with ModMed's product team on requirements, quality definition, and issue resolution
GlobalLogic
Sep 2017 – Jul 2018
Noida, India
Senior QA Engineer — Ignite TV (OTT Platform)

Nine months on Ignite TV — live TV, On Demand, Cloud PVR. OTT has no patience for flaky tests; if something slips through, users notice immediately. Owned the test strategy, improved release predictability, moved on better for it.

  • Owned test strategy and automation for Ignite TV — live TV, On Demand, and Cloud PVR features
  • Developed automated test scripts and improved release quality through scalable end-to-end validation
  • Improved release quality and execution predictability through cross-team alignment
NEC Technologies India
Nov 2015 – Sep 2017
Noida, India
Test Engineer — Equinox (Gym Management SaaS)

First time building automation from scratch rather than inheriting someone else's work. Selenium + Java on Grid, owned end to end — that kind of ownership teaches you faster than anything else does.

  • Built and maintained automated tests using Selenium + Java on Selenium Grid
  • Improved test coverage and execution efficiency by 45%
  • Introduced structured defect reporting and test coverage tracking across the team
Web View Solutions
Jul 2014 – Nov 2015
India
QA Engineer — Hotel Management System
  • Tested web-based HMS and POS system end-to-end — validated new features across the full hotel lifecycle automation solution
  • First engineering role — where the curiosity about why products fail began

Education & Certifications

Degree
B.Tech — Electronics & Communication Engineering
UPTU (now AKTU) · 2010 – 2014
Certificate
Professional Certificate in Product Management
ISB Hyderabad · 2025
Certifications
PMP® CSPO® ISTQB® Foundation

Projects

Case studies
in the making.

Real work. Real outcomes. Real numbers. Documenting 12 years of product decisions — from breaking software intentionally to shipping at 10M+ scale — into case studies that show the thinking, not just the result.

● Building

Get in touch.

Mobile +91 94568 29802
Email cvsingh99@gmail.com
LinkedIn linkedin.com/in/chandra-vijay-singh ↗
X / Twitter @cvsingh99 ↗