zainishtiaqahmad@gmail.com

Hi, my name is

Zain Ishtiaq.

I build systems that scale to millions.

I'm a senior software engineer with 11+ years of experience building high-scale distributed systems, fintech platforms, and AI-powered automation. Currently at Fastn, where I've shipped 2,400+ commits, own 2 of 5 backend services, and built the AI agent runtime from scratch.

Get In Touch

About Me

I specialize in creating scalable software solutions and leading projects involving Java/Spring, Mobile App Development, Microservices, MERN Stack, and DevOps practices. I excel in performance enhancement, production support, and agile team collaboration, including leading peers and mentoring team members.

My journey started at Techverx building healthcare and e-commerce systems, then I spent nearly 5 years at Careem (acquired by Uber) architecting backend systems for ride-hailing and the CPay fintech platform serving millions of users. Today at Fastn, I'm the designated single owner of 2 of 5 backend services, built the AI agent runtime and multi-cloud KMS encryption from scratch, and serve as the de-facto release captain and lead incident commander.

Here are some technologies I work with:

  • Java / Spring Boot
  • JavaScript / TypeScript
  • Go
  • Kotlin
  • Vue.js / React
  • Node.js / MERN
  • Kafka / gRPC
  • Docker / Kubernetes
  • AWS / Azure / GCP
  • Microservices
  • PostgreSQL / MongoDB
  • Terraform / CI/CD
Zain Ishtiaq

Where I've Worked

Senior Full-Stack Engineer @ Fastn

October 2024 — Present

  • Authored 2,400+ commits across 14+ repositories and shipped two foundational subsystems — Quartz-based scheduling and the AI Agent runtime (OpenAI, Gemini, MCP integration)
  • Designated single owner of 2 of 5 backend services (Event Service and Workflow Service) in the v2 platform rewrite
  • Built the AI Cost Attribution platform end-to-end — ledger, 5 API endpoints, full admin UI with daily-spend charts, CSV export, and per-agent drill-down
  • Architected the SSE to Streamable-HTTP migration across all 10 agent endpoints, fixing GCP load-balancer "200s network error" with Redis-backed event buffering and Last-Event-ID resume
  • Designed multi-cloud KMS encryption (AWS / Azure / GCP) with per-tenant AES-256-GCM keys, replacing insecure AES-ECB
  • Built centralized observability — 14 Grafana alert rules, 4 operational runbooks, PagerDuty integration, and error metrics pipeline synced via GitHub Actions
  • Developed a custom alternative to AWS EventBridge and Pipes, creating an efficient event-driven architecture tailored for the platform
  • Worked closely with HP and Fabric teams to design and implement seamless enterprise integrations
  • Serve as de-facto release captain, lead incident commander, and primary mentor for the platform team

Senior Software Engineer @ Careem

December 2019 — September 2024

  • Led the design and development of scalable backend systems using JavaScript, Node.js, Java, and microservices architecture, catering to high-traffic applications
  • Built robust backend infrastructure with Go, Java, Spring Boot, and RESTful APIs, optimizing performance for large-scale user bases
  • Developed and implemented a gRPC-backed service for dynamic fare negotiation in ride-hailing, enhancing performance and scalability with the Flexi Rides feature
  • Implemented Kafka for real-time data processing, optimizing data streams and enhancing the overall system architecture
  • Played a key role in CPay, Careem's fintech platform, deploying the P2P payments module enabling secure and efficient peer-to-peer transactions
  • Utilized Docker and Kubernetes for containerization and deployment, ensuring seamless scalability and high reliability in production
  • Contributed to the development of modules with a user base in the millions, ensuring high performance and availability
  • Authored RFCs and PRDs, actively engaging in product and design discussions to influence the technical direction and strategy of the company
  • Onboarded and mentored dozens of new engineers, fostering their integration and contributing to professional growth

Software Engineer @ Techverx

November 2015 — October 2019

  • Led the deployment and optimization of RysaRx, a US-based pharmacy system handling patient, prescriber, and insurance billing microservices
  • Led the migration of legacy systems to a microservices architecture, improving scalability and maintainability
  • Implemented frontend modules using Vue.js for healthcare, workforce management, and e-commerce solutions
  • Spearheaded the development of backend and Android apps for My Worker and Moishe House, improving user engagement and platform functionality
  • Implemented RESTful APIs and integrated third-party services to enhance application features
  • Designed and optimized MySQL databases for efficient data handling and management
  • Managed project timelines, defined project scope, and provided technical guidance while mentoring junior developers

Featured Work

2,400+
Commits Shipped
480+
PRs Merged
290K+
Lines of Code
14+
Repos Contributed
Fastn v1 + v2

AI Agent Runtime

Built fastn's entire AI agent capability from scratch — OpenAI & Gemini LLM clients, session memory, MCP tool integration, prompt builder (41 commits), and the AI Cost Attribution platform with full admin UI, daily-spend charts, and per-agent drill-down.

Java· TypeScript· LangChain· Vercel AI SDK· MCP
🔒 Fastn v1

Multi-Cloud KMS Encryption

Designed EncryptionStrategy interface with AWS KMS, Azure Key Vault, GCP Cloud KMS, and Local strategies. Replaced insecure AES-ECB with per-tenant AES-256-GCM using random IVs and v2: versioned ciphertext.

Java· AWS KMS· Azure Key Vault· GCP KMS
Fastn v1

Quartz Scheduler Subsystem

Entire scheduling subsystem authored from scratch — QuartzSchedulerService, JobInitializer, listeners, and ScheduleUtils. Later led a multi-week reliability rescue with health checks, connection-pool tuning, auto-recovery, and PagerDuty alerting.

Java· Micronaut· Quartz· Grafana· PagerDuty
🚀 Fastn v2

Streamable-HTTP Migration

Migrated all 10 agent endpoints from SSE to Streamable-HTTP with Redis-backed event buffer, monotonic event IDs, Last-Event-ID resume, and close-race-safe terminal frames — fixing GCP load-balancer "200s network error" on long streams.

TypeScript· Redis Streams· GKE· Express
📈 Fastn v2

Observability & Alerting Platform

Built centralized error metrics pipeline with ERROR_REGISTRY of 28 codes, auto-emitting fastn_errors_total on every 4xx/5xx. Authored 14 Grafana alert rules, 4 runbooks, 2 dashboards, and a GitHub Action that syncs YAML to Grafana Cloud.

Prometheus· Grafana· GitHub Actions· YAML
💰 Careem

CPay Fintech Platform

Key contributor to Careem's fintech platform serving millions of users — deployed the P2P payments module, built gRPC-backed fare negotiation service for Flexi Rides, and implemented Kafka for real-time data processing across ride-hailing services.

Node.js· Java· Go· gRPC· Kafka

Stuck-Stream "I'm Done!" Race Condition

Final stream frame was saved server-side but never delivered, leaving dashboard agent chats hanging forever. Introduced close-race-safe terminal frames in createStreamableSession and rolled out across all 10 agent endpoints. 10,000+ lines changed across 4 PRs.

Fastn v2 · Streaming · Redis

JWT Trust Boundary Regression

Gateway was trusting JWT-claimed org_id and role from Keycloak without verifying membership, allowing potential cross-account access. Reworked to resolve org + role from verified account_memberships join and scoped impersonated sessions at the gateway layer.

Fastn v2 · Security · Auth

Production OOM Kills — Thread Pool Bottleneck

Identified Micronaut's default 2-thread scheduled executor causing a 3,447-task backlog leading to OOM kills. Traced through @Async queue backlog and shipped a configuration-only fix bumping core-pool-size to 50 across all environments.

Fastn v1 · JVM · Performance

Upstream LLM 402 Leaking Customer-Facing Copy

When Vercel AI Gateway returned 402 (exhausted credits), the agent service streamed "Insufficient funds" verbatim to customers, implying they were out of credits. Masked all upstream LLM error copy as PROVIDER_UNAVAILABLE with generic UX.

Fastn v2 · AI · UX

Cin7 Core 503 — 16KB Body Limit

readBody() defaulted to 16KB max. The Cin7 Core "Create product" action had 72 input fields — the JSON payload exceeded the limit, causing req.destroy() and Envoy 503s. Traced through raw logs and raised the limit to 1MB to match existing routes.

Fastn v2 · Debugging · Infra
What's Next?

Get In Touch

I'm currently open to new opportunities and my inbox is always open. Whether you have a question, a project idea, or just want to say hi — I'll do my best to get back to you.

Say Hello