Skip to content

Available for freelance & remote work

I build reliable agent systems.

AI engineer specialising in agentic automation, voice agents and RAG — plus the evaluation and guardrails that keep them working in production.

Work with meSee the work

Currently AI Engineer at Joblogic · MS Artificial Intelligence, LUMS

LUMSMS Artificial Intelligence · 3.82 GPA · full scholarship
Gold MedalBS Maths + CS, GCU Lahore
~4 yrsproduction AI & software engineering

Selected work

Systems shipped, not slides.

Four builds — three at Joblogic, one for a paying freelance client. Each links to a full case study with the architecture and the hard parts.

All work
FlagshipJoblogic

Automation Design Portal

An agentic platform that lets automation engineers build new integrations in natural language. A LangGraph agent discovers API endpoints, studies related repositories, learns the codebase's conventions, gathers clarifications, and writes a detailed PRD — which Claude Code then implements and tests.

  • LangGraph
  • Claude (Opus 4.8)
  • RAG over code
  • Python
Private · walkthrough on requestRead case study

AI product · paid client

Freelance

AI Job-Matching & Application Platform

Aggregates job listings across platforms, builds a profile from an uploaded CV with GPT-based parsing, then scores relevance, summarises each role, analyses skill matches and gaps, and generates tailored cover letters and applications — with notifications for high-match jobs and incomplete applications.

  • GPT parsing
  • Matching & scoring
  • Next.js
  • Automation
Paid clientCase study

No-code agent builder

Joblogic

Agent Design Portal

An internal, Agentforce-style no-code agent builder. A builder chatbot that knows the company's available tools lets users describe an agent in conversation; the platform then assembles, configures and deploys it — with triggers across email, voice calls and WhatsApp.

  • Multi-agent orchestration
  • ElevenLabs (voice)
  • Nylas (email)
  • WhatsApp
  • LLM tool-calling
Private · walkthrough on requestCase study

Multi-modal data → insight

Joblogic

Customer Data Analysis Platform

A pipeline turning messy customer-communication archives into business insight. Tenants upload ZIP/PST/MSG; the pipeline lands them in Databricks, normalises and transcribes audio, embeds and cleans text, reduces with UMAP and clusters with HDBSCAN, then sends representative conversations to Claude to surface use-cases and interaction patterns.

  • Databricks
  • Transcription
  • UMAP
  • HDBSCAN
  • Embeddings
  • Claude (Sonnet 4.6)
Private · walkthrough on requestCase study

Currently building

What's next.

Open-source and product work in progress — the tools I wish existed for building and trusting agents.

Voice AI

Coming soon

Vertical Voice AI Agent

Booking, lead qualification and FAQs over the phone, with calendar/CRM sync and a call-outcomes dashboard.

  • Voice AI
  • ElevenLabs
  • Vapi / Retell
  • Twilio
In progressCall the demo line — soon

Infrastructure

Coming soon

Governed MCP Server / Gateway

A production Model Context Protocol server/gateway adding auth, per-tool authorization, rate-limiting, usage metering and audit logging. Open-source.

  • MCP
  • Infra
  • Governance
  • Open-source
In progress

LLMOps

Coming soon

Agent Reliability & Eval Suite

Hallucination/faithfulness plus cost/latency scoring with proper statistical confidence intervals, and a CI action that fails PRs on regressions. Open-source + hosted demo.

  • Evals
  • Observability
  • LLMOps
  • Open-source
In progress

What I do

Hire me to build the hard parts.

I work with founders and teams shipping LLM products — from a first agent to the evaluation and guardrails that keep it reliable.

Start a project

Agentic automation

Multi-step workflows an LLM agent plans and executes end to end — collapsing manual processes into a guided pipeline.

  • LangGraph & multi-agent orchestration
  • Tool-calling and API integration
  • RAG over your code, docs and data

Voice AI agents

Inbound and outbound voice agents that book, qualify leads and answer questions — wired into your calendar and CRM.

  • ElevenLabs · Vapi · Retell + Twilio
  • Calendar & CRM sync
  • Call-outcome analytics

AI reliability & evals

The tracing, evaluation and guardrails that turn a flaky demo into something you can trust in production.

  • Faithfulness & hallucination scoring
  • Regression gating in CI
  • Cost & latency observability

RAG & data pipelines

Turning messy, multi-modal archives into retrieval and insight at scale — the groundwork good agents stand on.

  • Embeddings & vector search
  • Clustering & topic discovery
  • Transcription & extraction

About

Engineering judgement, not just prompts.

I’m Hanzla — an AI engineer in Lahore, Pakistan, working with clients worldwide. For the past few years I’ve built production AI: agentic platforms that write their own integrations, no-code agent builders, voice agents, and the pipelines that feed them.

I care most about the unglamorous part — the evals, guardrails and tracing that decide whether an agent survives contact with real users. That’s the difference between a demo and a system you can sell.

More about me

Credentials

MS Artificial Intelligence

2024 – 2026

LUMS

CGPA 3.82/4.0 · Full merit scholarship · Top-5 cohort

BS Mathematics with Computer Science

Graduated

GCU Lahore

Gold Medal · Top of cohort