Eight years of shipping taught me that the hard part is never the framework. It's understanding what someone actually needs, then building something that survives contact with real users. That applies to AI systems more than anything I've worked on before.
the session, the stack and the links, as text ›
$ whoami
staff software engineer @ sastaticket.pk
builds agents. argues about consciousness.
$ cat now.md
an agent that analyses every support call and scores it
python · langchain + langgraph · gemini · claude · aws · gcp
the model is the easy part.
$ history | tail -5
edtech · erp/hrms · data security · travel · ai agents
us · europe · middle east · south asia
$ open /contact
got an interesting problem? let's build something.AI LangChain + LangGraph • Gemini • Claude • GPT • evals Languages JavaScript • TypeScript • Python • SQL Frontend React • Vue • Angular Backend Node.js • Django • GraphQL Data PostgreSQL • MongoDB Infra Docker • Kubernetes • Airflow • AWS • GCP • Azure • Sentry
Most of what I've built runs in production and isn't on GitHub —
resume is where it lives.
Curious about most things. Employed for some of them.