Engineering Lead|Lahore, PKT|Open to Senior / Tech Lead roles

I lead a team of 7 to 10 engineers and build the KYC, AML and identity systems used in 150+ countries.

Portrait of Syed Fakhar Abbas
Syed Fakhar AbbasEngineering LeadLeft, the photograph. Right, a 3D model of it built from about 22,000 points.

The KYC, KYB and AML platform I build, in numbers

  • 1M+monthly active usersPeople who verify their identity on it each month.
  • 150+countriesThe spread of those users across the world.
  • 60%less processing timePer verification, from rule engines and parallel API calls.
  • 99.9%uptimeThe uptime SLA for the platform.

// Approach

Verification is a chain of small decisions. I build the chain: rules, models and parallel orchestration, working as one pipeline.

01 / Live demo

Watch it work

A synthetic ID card goes through the five stages of a verification. Press Play, step through, or force an edge case to see it route to a human reviewer.

SPECIMEN REPUBLIC | IDENTITY CARDSPECIMEN, ALEX Q.01 JAN 1990SPEC-000000EXP 31 DEC 2030P<SPC<SPECIMEN<<ALEX<Q<<<<<<<<<<SPECIMEN

// STAGE 01 | CAPTURE

Capture and quality checks

Results from the platform this pipeline illustrates

  • 60%faster verification processingAutomated rule engines and parallel API orchestration.
  • 40%lift in user conversionFrom load-time and UX work in the verification journey.
  • 8+client-specific compliance configurationsDelivered with zero cross-client regressions.

Illustrative. A pipeline with synthetic data. The card is a generated specimen, no real document or face is processed, and nothing leaves your browser. Scores and thresholds are toy values for this demo, not the production rule engine.

02 / Live demo

Not a bot

Move your mouse across the pad and type in the box. Timing signals decide whether the input looks human. Then press Simulate bot and watch the score flip.

--Collecting signals
Pointer speed
Time between key presses
Pointer samples
0
Speed variation
0.00
Path turning (rad)
0.000
Path deviation
0.000
Key presses
0
Key timing variation
0.00
Key timing entropy
0.00

Illustrative. A simple heuristic running only in your browser. Nothing is recorded, stored or sent. Only pointer positions and the time between key presses are kept, in memory, and cleared on reload. Which keys you press is never stored. The production model behind the 98% human-vs-bot figure is far more involved and is not shown here.

03 / Live demo

How the work gets done

Pick a ticket. A command resolves, the runtime plans it, specialist agents do the work, and contexts feed them what they need. Different tickets fire different agents, because this is a designed system and not one chatbot.

// Pick a ticket
// Layer 01 | CommandsWhat was asked for
  • fix-bug
  • add-provider
  • patch-vulnerability
  • customise-flow
  • audit-pipeline
// Layer 02 | RuntimePlans, schedules and reports
  • Plan
  • Schedule
  • Report
// Layer 03 | AgentsSpecialists that do the work
  • Code
  • Test
  • Review
  • Security
  • OCR debug
  • DevOps
  • Integration
// Layer 04 | ContextsWhat the agents know and talk to
  • Ticket (ClickUp)
  • Repo knowledge
  • Team conventions
  • Source control (GitLab)
  • Team chat (Slack)

// run_log (generic, scripted)

    Runs on local models (Ollama, LLaMA, Mistral) or hosted ones, and is integrated with ClickUp, GitLab and Slack for ticket-driven orchestration.

    • 10+repositoriesOne orchestrated platform instead of several separate agent solutions.
    • 8+workflow typesFeature work, bug fixing, regression testing, DevOps audits, OCR debugging, client customisations, vulnerability patching, provider integrations.
    • About 30%less manual review overheadFrom prompt-chaining pipelines on local models.

    Illustrative. A simplified model of the architecture with generic tickets and a scripted timeline. No internal repositories, tools, prompts or customer data are shown, and nothing leaves your browser.

    04 / Projects

    Selected projects

    The three live demos above cover verification, bot detection and the agent platform. These are the other projects, with the numbers where I have them. Internals and customer data are confidential and not shown.

    • Dashboard

      Merchant cockpit

      100+languages through i18n

      A multi-language dashboard for merchants: real-time transaction monitoring, risk scoring, document review and case management, with configurable templates, webhooks and role-based access.

      • Vue 3
      • Pinia
      • Nuxt.js
      • RBAC
    • Product lead

      Facial recognition attendance

      99%face match accuracy

      1:1 and 1:N face matching at 99% accuracy, with active and passive liveness detection, responses in under a second and defences against 56+ spoofing types. Scaled to 10,000+ daily users.

      • Biometrics
      • Liveness
      • Face matching
    • Security

      KYC security hardening

      Zerocritical incidents since

      Closed injection attack vectors in live KYC stream processing with input sanitisation, query parameterisation and runtime validation, on a distributed multi-server architecture, with no downtime.

      • Input sanitisation
      • Runtime validation
      • Distributed
    • Machine learning data

      Behavioral WebSDK

      98%human vs bot accuracy

      A WebSDK that collects keystroke, mouse and touch signals in real time to train a model that tells humans from bots.

      • WebSDK
      • Behavioral signals
      • ML
    • Internal platform

      Manage-It

      80%lower cost, $25 to $5 per user per year

      An internal task management platform built as a ClickUp alternative, in Vue.js, Node.js and PostgreSQL.

      • Vue.js
      • Node.js
      • PostgreSQL
    • Tooling

      AI-assisted developer workflow

      About 30%less manual review overhead

      Local LLMs (Ollama with LLaMA or Mistral) chained with prompts for code review, test generation and documentation.

      • Ollama
      • LLaMA
      • Mistral

    05 / Experience

    Experience, 6 years drawn to scale

    Engineering Lead at Programmers Force for nearly 4 years, promoted from Associate Software Engineer in 26 months. Freelance work since Oct 2020 ran alongside. Bars are sized by real duration; select one to read it.

    Programmers Force

    Freelance

    Earlier

    Programmers Force

    Jan 2025 to now

    1 year 9 months so far

    Engineering Lead

    Team and standardsOwn how a team builds, not only what it builds.

    • Leads a team of 7 to 10 engineers.
    • Unified separate agent solutions into one orchestrated platform across 10+ repositories.
    • Led the platform decentralisation initiative.
    • Introduced PR templates, architecture decision records and review SLAs.

    Programmers Force

    Feb 2023 to Dec 2024

    1 year 11 months

    Full-Stack Engineer, KYC and AML platform

    SystemsOwn whole modules and how they behave under load and across clients.

    • Built the core KYC, AML, EDD, OCR, KYI and KYT modules.
    • Cut verification processing time by 60% with automated rule engines and parallel API orchestration.
    • Lifted user conversion by 40% through load-time and UX work.
    • Delivered 8+ client-specific compliance configurations with zero cross-client regressions.

    Programmers Force

    Nov 2022 to Jan 2023

    3 months

    Associate Software Engineer

    CodeOwn the feature in front of me.

    • Shipped production features on the KYC platform.

    Fiverr

    Oct 2020 to now

    6 years so far

    Freelance Full-Stack Developer

    Client work, alongside full-time roles.

    • 20+ client projects across fintech, SaaS and e-commerce, using MERN and Vue.js with Node.js.
    • Scoped, designed, built and shipped each one end to end, and handled the client conversation.

    i-Sight

    Mar 2022 to Sep 2022

    7 months

    Chief Executive Officer

    Self-employed, before joining Programmers Force.

    Human Centered Robotics Lab

    Aug 2021 to Sep 2021

    2 months

    Data acquisition intern

    Internship during the degree.

    06 / Skills

    From the browser to the model

    Vue.js hands-on for 4.5+ years. Each group below is something the platform needed: interfaces, services, identity, data, delivery, AI and compliance.

    Top skills

    • System design
    • Backend development
    • JavaScript
    • Microservices architecture
    • Python
    See every skill, grouped

    Languages

    • JavaScript (ES2022+)
    • TypeScript
    • Python
    • Go

    Frontend

    • Vue.js (Vue 2 and 3)
    • React.js
    • Next.js
    • Nuxt.js
    • Pinia
    • Vuex
    • Vue Router
    • Tailwind CSS
    • Vite
    • Webpack

    Backend

    • Node.js
    • Express.js
    • REST APIs
    • GraphQL
    • WebSockets
    • Microservices
    • System design

    Identity and auth

    • OAuth 2.0
    • OpenID Connect
    • CIAM
    • RBAC
    • Webhooks

    Data

    • PostgreSQL
    • MongoDB
    • Redis

    DevOps

    • Docker
    • CI/CD
    • GitHub Actions
    • Linux
    • Git

    AI and automation

    • Agentic workflow design
    • Local LLMs (Ollama, LLaMA, Mistral)
    • OCR pipelines
    • Biometric ML
    • Liveness detection
    • Risk scoring

    Compliance and security

    • KYC
    • KYB
    • AML
    • EDD
    • GDPR
    • FATF
    • PSD2
    • Fraud detection
    • Injection attack hardening
    • Audit logging

    Testing

    • Jest
    • Vitest
    • Cypress
    • Playwright
    • Mocha
    • Selenium

    07 / Education

    Trained as a control engineer

    Degree

    B.Sc. Mechatronics and Control Engineering

    UET Lahore

    It gave me the habit of asking about inputs, feedback and failure modes before writing code, which is the same way I look at a verification pipeline or an agent platform.

    Courses and assessments

    • Crash Course on PythonCoursera, Dec 2022
    • Using Python to Interact with the Operating SystemCoursera, Jan 2023
    • JavaScriptLinkedIn Skill Assessment, passed

    08 / Contact

    Hiring for identity, compliance or platform engineering? Tell me the problem.

    Contact

    STATUS
    Open to Senior / Tech Lead roles
    LOCATION
    Lahore, Pakistan, PKT, UTC+5
    EMAIL
    afakhar363@gmail.com
    ELSEWHERE
    LinkedIn and GitHub, links below