Skip to content

AI-First Software Engineer · Islamabad, Pakistan

I build the orchestration layers that let autonomous agents run unattended in production.

6+ years of backend and cloud engineering, now spent on agentic systems at Trilogy. Dual certified AWS Solutions Architect and GCP Professional Cloud Architect. The work I am proudest of is usually the layer underneath the thing people see.

Read the case studiesResumeBook 30 minutes
Portrait of Shahid Hameed Chaudhary
autonomous agents in production
130+autonomous agents in production
execution reliability
99.9%execution reliability
records under management
230M+records under management
client cloud savings identified
$20M+client cloud savings identified

What I do

Four things, in the order I would defend them

Ordered by where I spend my time now, not by how long I have been doing each one.

Agentic AI and LLM Systems

Designing the operating model around autonomous agents, not just the agents themselves.

130+ autonomous agents in production at 99.9% execution reliability, with a failure rate under 0.1%.

  • Multi-step agent orchestration with retries, dead-letter queues, and structured output validation
  • Execution policies, tool and service access scoping, and leadership-visible observability
  • Model-agnostic integration and routing across providers, with cost and latency policy enforcement
  • LLM-driven classification, natural-language generation, and automated validation in the loop
  • Agent-first development workflow built on custom agents, skills, and scoped MCP servers
AWS Step FunctionsAWS BedrockOpenAIAnthropicPortkeyTrueFoundryMCPClaude Code

Cloud Architecture and Serverless

Dual certified on AWS and GCP, with the production scars to go with the badges.

$20M+ in client cloud savings identified, and a 35% Lambda cost reduction across 40+ functions at p99 under 500ms.

  • Event-driven and serverless architecture at millions of daily executions
  • Container orchestration and migration off serverless when the connection math stops working
  • Infrastructure as code, blue-green deploys, health checks, and auto-scaling
  • Cost optimization as an engineering discipline rather than a billing exercise
  • Multi-tenant systems with active replication and granular RBAC
AWSGCPAzureLambdaECS FargateDockerKubernetesTerraformCloudFormationGitHub Actions

Data and Streaming at Scale

The systems that look boring from outside and are the hardest thing in the building.

230M+ active records across 11 related tables, and query times taken from 15 minutes to milliseconds.

  • Real-time ETL pipelines with sub-200ms ingestion latency
  • Partitioning, indexing, and schema design decided before the first line of code
  • Connection pooling and database ceilings under serverless fan-out
  • Data lake and warehouse design with Parquet, S3, and BigQuery
  • Zero-loss bulk migration and corrupted database recovery
Apache KafkaAWS KinesisPostgreSQLMongoDBMySQLRedisDynamoDBBigQueryParquet

Full-Stack Engineering

Six years of shipping the whole thing, from schema to screen.

Led a 10-person cross-functional pod to zero production incidents, with PR turnaround under 4 hours.

  • REST and GraphQL API design in Node, Python, and Go
  • Production React and Next.js front ends with an eye on Core Web Vitals
  • Mobile delivery with Flutter alongside the web surface
  • Technical leadership: architecture review, delivery predictability, and mentoring
  • Security work as standard practice, OWASP remediation, encryption at rest and in transit, webhook verification
TypeScriptPythonGoReactNext.jsNode.jsNestJSDjangoFlaskFlutter

Experience

Where the scars came from

Five engagements, each a different way for a system to fall over at volume. The detail lives in the case studies.

AI-First Software Engineer

Trilogy · Enterprise cloud cost optimization SaaS

Jul 2025 to Present · Islamabad, Pakistan · Remote

Build and operate the platform that lets 130+ autonomous agents run unattended against customer AWS estates and be trusted with the result. Less about any one agent than about the orchestration, validation, and policy layer underneath all of them.

Autonomous agents
130+
Execution reliability
99.9%
Client savings identified
$20M+
AWS Step FunctionsLambdaECS FargatePythonTypeScriptBedrockPortkey
Read the case study

Senior Software Engineer, Platform

OptoCeutics · Digital health and neurotech platform

May 2024 to Jul 2025 · Copenhagen, Denmark · Remote

Led platform engineering and the pod that shipped it, for a product serving 50K+ users. Inherited five legacy systems and a sprawling Lambda estate, and spent the year making it something a small team could operate.

Users served
50K+
Infra cost reduction
40%
Pod size led
10
AWS LambdaApache KafkaBigQueryReactDjangoNestJSFlutter
Read the case study

Full Stack Developer

Techception · Interactive video wall SaaS

Feb 2023 to Mar 2024 · Islamabad, Pakistan · Hybrid

Scaled the platform past 165M requests and kept it there. The defining problem was a serverless architecture whose database connections grew in lockstep with traffic, until the database became the ceiling.

Requests served
165M+
Uptime
99.95%
Mongo connections
1,800 to 400
Node.jsReactMongoDBAzure FunctionsDocker
Read the case study

Software Engineer II

Elite IT Team · US real estate lead platform

Nov 2020 to Feb 2023 · Islamabad, Pakistan · Remote

Owned the backend of a lead distribution platform holding 230M+ records across 11 related tables, and led the backend of two SaaS products on top of it. At that size the decision made before any code is written is the one that matters.

Active leads
230M+
Query time
15 min to ms
Monthly active users
10K+
Node.jsPythonPostgreSQLRedisECS FargateDocker
Read the case study

Backend Developer

Vintage Solutions · Ride-hailing platform, UAE

Feb 2020 to Nov 2020 · Remote · Remote

First production backend work. Real-time distance and geofencing logic supporting 1,000+ daily bookings, across a polyglot data layer and RBAC over three microservices.

Daily bookings
1,000+
Node.jsMySQLMongoDBPostgreSQL

Case studies

The problems worth writing up

Each one covers the constraint, what it cost, and what I would do differently.

All case studies

Writing

Notes from the trenches

Mostly infrastructure, mostly things that surprised me. Several started as LinkedIn posts and grew.

All writing

Lab

Things I rebuilt to understand them

Public repositories, mostly infrastructure primitives taken apart to see how they work. Reading the paper is not the same as running the thing.

  • explore-firecrackers

    Shell

    Benchmarked Firecracker microVM boot times to see whether VM-grade isolation can run at container speed. It can: 216.7ms average boot, under 2% variation.

    RepositoryWrite-up

  • explore-ratelimiter

    Go

    Rate limiting algorithms implemented from scratch in Go, compared under load rather than on a whiteboard.

    Repository

  • Redigo

    Go

    A Redis-inspired key-value store in Go. Written to understand the protocol and the event loop by rebuilding them.

    Repository

  • explore-prometheus

    TypeScript

    Instrumentation and metrics pipeline exploration, focused on what is actually worth alerting on.

    Repository

  • RAG-Demo-Suite

    Python

    Retrieval-augmented generation patterns compared side by side, including chunking and retrieval tradeoffs.

    Repository

  • explore-devcontainers

    Shell

    Dev container configurations for reproducible environments across projects and machines.

    Repository

Shipped

Products with something to point at

Client and product work where there is a live URL rather than an architecture diagram.

  • Reprosify

    Backend lead

    Realtor platform built on top of 230M+ leads, with distribution to thousands of agents and query latency measured in milliseconds.

    Node.jsPostgreSQLRedisNext.jsTailwind CSS

  • Cuisinette

    Full-stack

    Multi-restaurant cloud kitchen ordering platform, from menu management through to dispatch.

    Next.jsNode.jsPostgreSQL
  • Anread screenshot

    Anread

    Full-stack

    Tech, policy, and development consultancy site with a content portal behind it.

    ReactNext.js
  • Reprosify Agents screenshot

    Reprosify Agents

    Full-stack

    Agent listing platform and acquisition funnel wired directly into the CRM.

    Next.jsNode.jsCRM integration

Credentials

Verifiable, not just claimed

Every badge below links to the issuer. The degree links to the university registry.

Certifications

  • Google Cloud Professional Cloud Architect

    Designing scalable, secure, and cost-effective architectures on Google Cloud.

    Google Cloud · Nov 2024 to Nov 2026

    Verify on Credly

  • AWS Certified Solutions Architect, Associate

    Designing and deploying secure, high-performing systems on AWS.

    Amazon Web Services · 2024

    Verify on Credly

Contact

Keep in touch

Based in Islamabad, Pakistan, working PKT, UTC+5. Comfortable overlapping with European mornings and US mornings, and used to fully remote teams.