class HarisAhmed:
role = "Associate AI Engineer @ Vanar"
degree = "BS Computer Science — NUML, Lahore"
based_in = "Lahore, Pakistan 🇵🇰"
def what_i_do(self):
return [
"Ship production-grade LLM apps, RAG systems & SaaS platforms",
"Design composable multi-agent frameworks (typed ports, zero-dep cores)",
"Make AI outputs earn trust — adversarial verification, real citations",
"Own products end-to-end: architecture → backend → a UI people enjoy",
]
def ask_me_about(self):
return ["RAG", "agent orchestration", "LLM juries", "billing that actually works"]
def currently(self):
return "Watching LLM societies evolve, one tick at a time 🧫"- 🎉 Merged PR #68979 in freeCodeCamp/freeCodeCamp
|
Composable AI Agent Framework Agents snap together like USB devices — regardless of LLM, language, or host. Typed input/output ports, with automatic transforms quietly dropped in when two ports almost match. The core is 100% dependency-free: any LangChain, CrewAI, or plain Python agent wraps in a single line via duck-typed bridges.
|
Multi-Agent Emergent Behavior Platform A petri dish for LLM societies, built on Weaveflow. Spin up autonomous citizen agents with private goals and memory, drop them into rule-governed games, and watch cooperation, deception, and market efficiency emerge as real numbers, not anecdotes. Fully reproducible from a seed — the offline backend needs zero API keys.
|
|
Hallucination-Resistant Research Engine Research that has to earn its conclusions. Web searchers and deep readers fan out, then every fact stands trial before a panel of skeptics — a deliberately mixed jury of Claude, DeepSeek, Gemini & GPT — and a majority vote kills any claim they can't defend. Word-for-word source quotes, SSRF guards, and prompt-injection sanitization built in.
|
The infrastructure behind the AI Public API gateway with rate limiting and auto-generated OpenAPI docs · tiered Free / Basic / Pro access engine · Stripe credit billing metering hundreds of users in real time · five scraping engines unified behind one REST API, cutting costs ~40% · conversational AI personas replying in under 300ms.
|
-4C1D95?style=for-the-badge){kind=icon}
Open to collaborating on agent frameworks, RAG systems & AI tooling —
or just talking shop about making LLMs behave in production.
