Skip to content
Khizar Amin
Portrait of Khizar Amin

Khizar Amin

Forward Deployed Engineer (Backend & AI) at PackageX

I build backend and AI systems that seamlessly connect the tools companies already run.

Forward Deployed Engineer at PackageX. I sit with client operations teams, map how their work actually happens, and ship it as workflows wired into their ERP, storefront, delivery and identity systems.

Before that I spent two years building the platform's backend those workflows run on, and two years as lead backend engineer at a logistics startup.

Lahore, PakistanOpen to remote roles and relocation

Experience

  1. PackageX

    Islamabad, Pakistan

    Logistics and inventory platform. Clients keep their ERP and storefront; PackageX connects them and runs the warehouse floor.

    1. Forward Deployed Engineer (Backend & AI)

      Apr 2026 – Present

      • Map client operations into PackageX Workflows (modular, block-based flows) and ship them for $250K+ ARR accounts.
      • Built the integration layer and 20 production workflows for a multi-location fresh pet-food brand: NetSuite, Shopify, Onfleet and Ordergroove behind proxy workflows that own auth, caching and idempotency.
      • Designed and shipped AI-assisted inbound receiving for a residential care provider: a label photo is read by vision inference and routed to a PO, a person or an exceptions queue, backed by Entra ID, EMR and Business Central syncs.
      • Built a remote orchestrator for Claude Code agents: parallel sessions in isolated git worktrees, a live terminal in the browser, steerable from a phone.
    2. Senior Node.js Developer

      Jan 2024 – Apr 2026

      • Built core SaaS APIs in Node.js and Prisma on MySQL: fulfillments, manifests, parcels, items, audits and cycle counts, documents, imports and batches.
      • Owned the asset management module end to end: lifecycle, assignment, maintenance and warranty states, imports, and lock-contention fixes on the write path.
      • Engineered lease-based Redis coordination, after the leases in Facebook's Memcache paper, for safe idempotent writes under high-concurrency MySQL workloads.
      • Built cloud printing APIs and the PDF template engine; made label-sheet generation about 3.5× faster by deduplicating and prefetching image fetches.
      • Connected NetSuite and legacy client systems (sales orders, partial fulfillments, tax, purchase orders) and maintained CI/CD with GitHub Actions.
  2. Bridgelinx

    Lahore, Pakistan

    Digital freight platform; the first Pakistani startup to raise a $10M seed round.

    1. Lead Backend Engineer

      Feb 2022 – Jan 2024

      • Led a team of two engineers building the backend for 9 web and mobile apps on Node.js, Apollo GraphQL and Neo4j.
      • Migrated more than 95% of legacy, unstructured Firebase data into a unified Neo4j/GraphQL schema with minimal loss.
      • Ran AWS infrastructure and CI/CD: CloudFormation stacks, push-to-deploy pipelines from GitHub, EC2, S3, Lambda and SQS. Cut deployment overhead by 50%.
    • Node.js
    • GraphQL
    • Neo4j
    • Redis
    • AWS
    • CloudFormation
  3. Eastman Enterprises

    Lahore, Pakistan

    1. Software Engineer

      Aug 2021 – Jan 2022

      • Built backend APIs and managed third-party integrations.
    • Node.js
  4. LUMS

    Lahore, Pakistan

    Research with Dr. Ihsan Ayyub Qazi and Dr. Zartash Uzmi.

    1. Research Assistant

      Mar 2019 – Aug 2021

      • Worked on Advention, a system that picks low-latency Tor circuits and serves relevant ads to Tor users.
      • Prototyped LasTor- and NavigaTor-style path selection for latency before settling on power-of-d-choices circuit selection.
      • Built measurement tooling to study how GDPR affected page load times.
    • Python
    • Tor
    • Measurement

Toolbox

Backend
Node.js · TypeScript · Go · Python · REST · GraphQL · Prisma
Data
MySQL · PostgreSQL · Neo4j · Redis · Elasticsearch
AI & agents
Claude API · Claude Code · LangGraph · LangChain · RedisVL semantic caching · Vision extraction · Multi-agent systems · Guardrails
Integrations
NetSuite (SuiteQL, REST) · Shopify Admin GraphQL · Business Central · Microsoft Graph / Entra ID · Onfleet · Ordergroove · OAuth 2.0 / JWT · SFTP
Infra
AWS (EC2, S3, Lambda, SQS, CloudFormation) · GCP Cloud Functions · GitHub Actions · Docker · Vercel · Sentry · Linux

Projects

  • 2024

    Raft consensus and a replicated key-value store

    Raft in Go from the paper: leader election, log replication, the current-term commit rule and persistent state, then a key-value service on top that survives crashes, split votes and network partitions.

    Course code (LUMS Distributed Systems), shared on request. Every test ran 10× with and without -race.

    • Go
    • Distributed systems
  • 2026

    Distributed Postgres

    Sharding playground: three Dockerized Postgres shards behind a Go routing proxy that hashes a shard key on a ring, co-locates each tenant's tables and broadcasts reference data to every shard.

    • Go
    • PostgreSQL
    • Docker
  • 2026

    Expenses

    A personal expense tracker's backend in Go, replacing a Google Sheet: JWT auth middleware, accounts, categories, multi-line transactions, DTO mapping and table-driven tests.

    • Go
    • REST
    • JWT
  • 2024

    Understanding the hidden data costs of YouTube Shorts

    Extends 'Uncovering the Hidden Data Costs of Mobile YouTube Video Ads' (WWW '24) to Shorts. Measured data use on rooted Android with mitmproxy (past SSL pinning) and OCR over 'stats for nerds': replayed Shorts re-fetch the whole video, and ad sequences add heavy overhead.

    • Python
    • mitmproxy
    • Tesseract OCR
    • Android
  • 2023

    4×4×4 tic-tac-toe with deep Q-learning

    Two self-play agents learn a 3D board with a neural Q-function plus a tactical override for immediate wins and blocks. One agent learned to set up forks, and we then trained it against the competition-winning DP agent.

    • Python
    • TensorFlow / Keras
    • Reinforcement learning

Also

Education

  • MS Computer Science, Lahore University of Management Sciences (LUMS)

    2022 – 2025

    • Thesis: cost-effective content moderation using LLMs, advised by Dr. Ihsan Ayyub Qazi and Dr. Zafar Qazi.
    • Merit scholarship: 60% fee waiver (2023–2024).
    • Coursework: Distributed Systems, Reinforcement Learning, Information Retrieval, Design & Analysis of Algorithms, Topics in Internet Research.
  • BS Computer Science, FAST National University (NUCES)

    2014 – 2019

    • President of SOFTEC 2018: led a 200+ person team to run Pakistan's largest IT event.

Off the clock

I'm the person who reads the changelog before the release notes. If something is new and shiny (a model that dropped this morning, a kernel flag, a tiling window manager) I have probably installed it, broken my machine with it, and written down how I fixed it.

Some of those write-ups ended up with more stars than my code. I run Arch, by the way.

Away from the keyboard I box for fun, then undo all of it with baking. My Basque cheesecake is the most reliable thing I ship: burnt on top, creamy inside, zero bug reports. The recipe is Joshua Weissman's; the deployment is mine.

Hiring for an FDE, solutions or senior backend role?

khizaramin95@gmail.com

Lahore, Pakistan · Open to remote roles and relocation. Or grab the résumé (fair warning: it links back here).