Abdullah Siddique
Senior AI Engineer | LangGraph, RAG & Production AI Systems
About
AI engineer with 7+ years across software engineering and cybersecurity, building and operating production AI systems: multi-agent LangGraph architectures, RAG pipelines, and the infrastructure that makes them enterprise-ready, including evaluation suites, human-approval controls, tenant isolation, and PII protection. Took OffSec KAI from concept to production and built OSCAR, an AI support agent that cut contextualized ticket volume by 40%. Leads products from ambiguous requirements through production and live support.
Upwork profile details
Profile snapshot: Sep 26, 2026
Availability: As Needed — Open to Offers
Languages: English — Native or Bilingual
Linked accounts: GitHub, Stack Overflow
Working style: Solution Oriented, Detail Oriented
Profile skills:LangChain · Retrieval Augmented Generation · Python · TypeScript · FastAPI · PostgreSQL · MongoDB · Kubernetes · Amazon Web Services · NestJS · Prompt Engineering · Next.js · React
Watch Upwork introduction videoCurrent contracts
AI Agent Consultant & Mentor — LangGraph, Observability & Evals
Sep 18, 2026 – Present · Hourly · $40.00/hr
Senior Python Developers for AI Model Review — Talent Store
Sep 24, 2026 – Present · Fixed price
Awards and Certifications
OffSec CORE Award
CORE Award postLangChain Certified Agent Engineer
Verify credentialLangChain · Issued Aug 28, 2026 · Expires Aug 28, 2028
A LangChain Certified Agent Engineer has demonstrated proficiency across the entire Agent Development Lifecycle using LangChain's open-source ecosystem (LangChain, LangGraph, Deep Agents) and LangSmith (Observability, Evaluation, and Deployment). LangChain verifies this proficiency through a two-hour, proctored exam built on five LangChain Academy courses.
Learning outcomes
OffSec Certified Professional (OSCP)
Verify credentialOffSec · Issued Aug 29, 2021 · No expiration
Demonstrated practical penetration-testing skills in information gathering, exploitation, privilege escalation, web attacks, and network pivoting.
Learning outcomes
OffSec Experienced Penetration Tester (OSEP)
Verify credentialOffSec · Issued May 27, 2022 · No expiration
Demonstrated the ability to test hardened systems, bypass security defenses, avoid detection, and compromise systems configured with security in mind.
Learning outcomes
Work Experience
The Be Human CompanyVancouver, BCSelf-employedFull Time
Co-Founder & CTO / Chief AI Officer
- Translated AI governance, security, and customer requirements into the product roadmap, system architecture, and delivery priorities for a governance-first SaaS platform.
- Built Python services for AI control evaluation and agent sandboxing with strict validation, automated tests, CI/CD, observability, and secure execution boundaries so customer workflows could be evaluated safely.
- Worked with technical and non-technical stakeholders to turn compliance evidence into maturity scores, prioritized remediation, and customer-facing reports.
- Built a fail-closed authorization gateway that verifies signed client authorization before agents can access real data, plus DMARC-based email trust validation, secrets-by-reference credentials, and network egress allowlists.
- Architected a Python AI governance platform with 65+ controls across eight domains, configurable maturity scoring, prioritized remediation, automated PDF evidence reporting, and an isolated synthetic-data sandbox with per-step tracing and nine pytest suites.
OffSecRemote - New York, USContractFull Time
Software Engineer L2 – AI Innovation & Development
- Built and operated Python and TypeScript/NestJS back-end services and versioned REST APIs for OffSec's SaaS cybersecurity-learning platform, enabling learners, enterprise administrators, and support teams to use AI guidance, recommendations, and reporting in production.
- Designed distributed and event-driven services with AWS, Kafka, PostgreSQL, MongoDB, Redis, and Elasticsearch, balancing latency, throughput, backward compatibility, and reliability as product requirements evolved.
- Owned services from development through production with automated testing, GitLab CI/CD, Docker, Kubernetes, Grafana, Loggly, OpenTelemetry, monitoring, alerting, and incident diagnosis so failures could be detected and resolved before disrupting customer workflows.
- Designed secure LLM and agent controls using authorization, account-scoped data and caches, read-only SQL, retries, audit logging, and PII-safe logs to protect sensitive enterprise and learner data.
- Collaborated in agile delivery with product managers, data scientists, designers, support, and engineers to translate ambiguous business needs into user-facing features; shared architecture guidance and mentored teammates through releases and production operations.
- Architected a LangGraph/Deep Agents B2B orchestrator that coordinated remote specialist agents across multiple microservices, strengthening enterprise retention and protecting tens of thousands of dollars in recurring revenue.
- Owned deployment and CI/CD for a 38-service platform across five development, preproduction, production, and blue/green environments using Docker Swarm, Traefik, Ansible, Vault-gated GitLab pipelines, and automatic rollback on failure.
JSW AI ServicesRemote - United StatesContractPart Time
Principal AI Engineer
- Architected multi-agent platforms for enterprise clients so teams could retrieve knowledge, analyze data, generate content, and execute approved tools through one workflow.
- Built JWT-secured typed and streaming APIs, React/Next.js interfaces, MongoDB retrieval, and LangSmith observability so authorized users could operate and evaluate agent workflows.
- Created evaluation datasets, pytest suites, pre-commit checks, CI/CD gates, and production feedback loops so agent releases could be tested before deployment.
- Developed Monet AI's LangGraph supervisor coordinating three specialist agents for MongoDB Atlas hybrid RAG, experience creation, and Snowflake analytics, with human approval gates, dual-JWT authentication, tenant isolation, and multi-turn LangSmith evaluation.
Polymer Runtime Data SecurityRemote - New York, USContractPart Time
Senior Python Developer
- Improved Python NLP detection for PII and PHI in Polymer's runtime data-security platform, tuning classification behavior to reduce false positives in customer DLP workflows.
- Optimized high-throughput remediation pipelines so security events were processed with lower latency.
- Extended DLP controls to AI traffic so customers could inspect and govern sensitive data entering or leaving AI systems.
- Integrated 10+ SaaS systems using OAuth and KMS-backed tenant isolation so customers could protect data across cloud applications.
- Worked with security and engineering teams under SOC 2 and HIPAA controls, mentoring engineers on safe handling of customer-sensitive data.
OffSecRemote - New York, USContractFull Time
Senior Student Mentor → Student Mentor – Technical Services
- Mentored OSCP, OSEP, and advanced cybersecurity learners through structured coaching, technical troubleshooting, documentation, and cross-functional escalation.
- Maintained production Discord bots and support automation so the Technical Services team could respond to learner questions faster and more consistently; monitored reliability and iterated from learner and support-team feedback.
- Automated exploitation and validation workflows to test lab and certification-exam infrastructure end to end, reproducing and escalating defects with engineering teams.
- Increased support efficiency threefold by developing and maintaining two production Discord bots for OffSec learners and technical support staff.
Pakistan Stock ExchangeKarachi, PakistanEmployeeFull Time
Data Vending Officer – Product & Research
- Delivered capital-market data products in a regulated exchange: index creation, maintenance, market analysis, investor reporting, and research on critical financial datasets.
- Maintained index methodologies and source-data quality, validating calculations, research inputs, and reporting outputs under exchange controls.
- Collaborated with product, research, data, and market stakeholders to translate requirements into investor-facing datasets, analysis, and reports.
Education
Lahore University of Management Sciences
Institute of Business Management
Skills
Programming languages
Frameworks & libraries
Databases
Tools & software
Cloud platforms
Soft skills
Compliance & security
Projects
Open source, awards & security projects
Oh My DCode
oh-my-claudecode's multi-agent orchestration layer, ported to the LangChain Deep Agents (Deep Agents Code) framework for TypeScript.
LangChain Skills
AnonymizerTS
TypeScript implementation of Microsoft Presidio using transformers.js for PII detection and anonymization.
TrustCallJS
Utilities for validated tool calling and extraction with retries using LLMs; a TypeScript port of trustcall.
Semantic Router
OffSec CORE Award
OffSec recognized Abdullah for work supporting the team and improving the learner experience.
Code Graph RAG
Autopwn Framework
Automation framework used to test and validate lab and certification-exam infrastructure.
OffSec Hints Bot
Discord bots providing graduated hints and support automation for OffSec learners and staff.
Oh My MuseCode
An OMC-shaped delivery harness for Meta Muse Code: deep-interview -> ralplan -> ralph, plus deep-dive/trace, team and cancel.
OMP TypeSafe
TypeSafe AI (Jev) adversarial reviewer and typesafe_ask tool for the omp coding agent.
Proprietary projects
Nymble AI Coach — Production Healthcare RAG Platform
Production healthcare coaching experience built around retrieval-augmented generation.
OffSec KAI
AI learning assistant built from concept to production for OffSec learners.
Monet AI
LangGraph supervisor coordinating knowledge retrieval, experience creation with human approval, and Snowflake analytics.
Slack QL Agent
Slack agent for plain-English questions against production MySQL, with read-only SQL, chart rendering, and query audit logs.
AI Governance Audit Framework
Governance platform with 65+ controls across eight domains, configurable maturity scoring, prioritized remediation, PDF reporting, and an isolated synthetic-data sandbox.
OffSec Recommendations Service
LangGraph and MCP service that turns team learning needs into training plans and lab recommendations.
OffSec AI Gateway
LangGraph-compatible API hosting the KAI tutoring graph with persistent checkpoints.
OffSec LangGraph Agent Server Toolkit
OffSec Reports Microservice – LangGraph Rewrite
OffSec Enterprise Learning Path & Preferences Agents
Cybersecurity Training
Heist Box Walkthrough
OffSec Live | Walkthrough of a PEN-200 AD Set
DigitalWorld.Local - BRAVERY -
Alice - PEN-200 Labs -
Press ⌘J to open the command menu