Senior full-stack product engineer building web, mobile, growth, and applied AI systems that survive the real world.
I like the messy middle: web apps, ERPs, video streaming products, mobile apps, product analytics, retention/conversion campaigns, user feedback, LLM quality, logs, QA automation, and the operating system around shipping useful products.
React.js · Next.js · React Native · Android · iOS · NestJS · Node.js · Firebase · AWS · MongoDB · Product Analytics · Growth Campaigns · Pinecone · Multi-agent orchestration · tmux · QA Automation · SFOS
| If you have 30 seconds | If you have 3 minutes | If you are technical |
|---|---|---|
| Visit my portfolio. | Read the WittyKeys case note. | Read the SFOS case note. |
I have spent 10 years shipping product software across web apps, ERPs, video-streaming platforms, mobile apps, backend systems, analytics, user-feedback loops, and growth campaigns. Recently I have been turning that senior product engineering base toward applied AI: not only model calls, but complete product systems around them.
My current body of work is centered on WittyKeys, a public Android AI keyboard and writing assistant on Google Play, plus a set of AI workflow systems around outreach, RAG support, campaign analytics, automated QA, evaluation, and my own solo-founder operating system.
The part I care about most: making AI products useful after the demo, when real users, weird states, billing, logs, regressions, and release decisions show up.
Before this AI chapter, I built and improved products through the full loop: collect engagement events, survey users, understand where people find value, plan sprints around that signal, run retention/conversion campaigns, and move the product in the direction users are actually choosing.
|
Live Android AI product. AI keyboard and writing assistant powered by Claude through Firebase. The hard parts were keyboard trust, overlay UX, onboarding, entitlement states, quota/paywalls, Play Billing, regression checks, observability, and Play Store release work. |
Human-reviewed outreach workflow. Campaign setup, prospect research, Claude-assisted drafts, regeneration, approval queues, follow-up, analytics, and browser E2E boundaries. Final sends and account-sensitive actions stay approval-gated. |
|
RAG support system design. Retrieval, citations, confidence scoring, escalation, knowledge-gap loops, and the product decisions behind useful support automation. |
Solo Founder Operating System. My operating loop for building WittyKeys with AI: Cowork planning, Claude Code implementation, worktrees, slash commands, golden screenshots, Espresso/UI Automator, LLM quality scoring, JourneyTracer, logs, and release gates. |
This is the part that feels most like me as an engineer.
idea -> plan -> build -> capture -> compare -> test -> score -> trace -> ship -> learn
| Layer | What I built around it |
|---|---|
| 🧑💻 AI-orchestrated development | Cowork plans, Claude Code implements, I review and approve gates. Parallel git worktrees let independent sessions work safely. |
| 🎨 Automated UI development | HTML/mockup references, structured implementation instructions, screenshot capture, visual comparison, issue triage, and fix loops. |
| ✅ QA automation | Golden screenshots, PixelDiffComparator, Espresso/UI Automator, real-device regression on Motorola Razr 50, and release checks. |
| 🧪 AI quality evaluation | LLM-as-judge scoring for reply quality, tone, Hinglish naturalness, context relevance, and output drift. |
| 🔭 Observability | JourneyTracer, sanitized traces, backend logs, Crashlytics, Firebase Performance, and post-release checks. |
I am looking for senior roles where product taste, full-stack execution, web/mobile depth, analytics, growth loops, and applied-AI judgment matter together:
- Senior Applied AI Engineer
- AI Product Engineer
- Forward Deployed Engineer
- Founding Engineer
- Senior Full-Stack Product Engineer
Looking for job opportunities.
Portfolio: wittykeys.com/abhishek · LinkedIn: abhishek-wittykeys