About
Darab Ahmed
I have been writing software since 2022, starting with an internship at Arbisoft. From 2023 I worked at Devsinc on Python, Django and React projects, and since July 2024 I have been at i2c on Java backend systems.
At i2c the code I write runs on a platform that banks and fintechs depend on, under PCI DSS. That is where I learned to measure before changing anything and to keep each change small.
Since AI coding tools arrived at work, most of what I build on the side is about making them useful on a large, private codebase: safe access to our systems, and the context they need to give good answers.
Experience
- i2c Inc. · Software EngineerJul 2024 – Present · 2 yr 4 mo
- Devsinc · Software EngineerApr 2024 – Jun 2024 · 3 mo
- Devsinc · Associate Software EngineerMar 2023 – Apr 2024 · 1 yr 2 mo
- Arbisoft · InternJul 2022 – Sep 2022 · 3 mo
– Present
Software Engineer, i2c Inc.
Lahore
Backend development
- Build and maintain backend modules in Java (Struts, Apache Tomcat, Informix) for an enterprise fintech platform used by banks and fintechs, under PCI DSS.
- Design the data flow before writing code, and take every change through code review and SonarQube, PMD, SpotBugs and Checkstyle.
- Write multi-threaded batch jobs for high-volume financial data, with audit logging for traceability.
- Took part in the platform’s migration from Java 8 to a current Java LTS.
Performance and production support
- Diagnose slow queries and jobs from execution plans and profiler output.
- Investigate production issues on live systems from logs and data, find the root cause and ship small, reviewed fixes, often the same day.
Tools for the engineering team
- Built Multi-Manager, a secure server-access tool that also gives Claude controlled access over MCP.
- Designed and built the Knowledge Compiler, a GraphRAG system that learns services in isolated containers and links about 30 of them into one knowledge graph.
–
Software Engineer, Devsinc
Lahore
- Built backend services in Python (Flask) with PostgreSQL.
- Containerised applications with Docker and deployed them to the cloud.
–
Associate Software Engineer, Devsinc
Lahore
- Migrated a legacy .NET application to Python (Django) with a redesigned schema and repeatable data migration scripts.
- Set up cloud infrastructure and CI/CD pipelines with GitHub Actions.
- Built frontend features with React and Next.js.
–
Intern, Arbisoft
Lahore
- Built a learning project with Python (Django) and a React frontend.
How I work
- 01
Performance work starts with evidence
Before changing a slow query I read its execution plan, and before changing slow code I profile it. The fix ships with before-and-after numbers.
- 02
Changes stay small and reviewed
In production support I look for the root cause first, then keep the fix small enough to review quickly and to roll back if needed.
- 03
Repeated work gets automated
When a task repeats and needs little judgment, I automate it: Multi-Manager for server access, and Claude Code skills for strict procedures such as DDL generation.
Example: a report query that got slow
A sample case of the kind of problem I look into at work.
- 1 · Symptom
- The statement report for a single account takes 38.4 seconds in QA.
- 2 · Diagnosis
- The plan shows a full scan of the transactions table. The filter wraps the indexed date column in a function, so no index can be used: every row is read, filtered, then sorted.
- 3 · Fix
- Filter on a range of the raw column, and add a composite index that matches both the filter and the sort order. Then measure again.
SQL
SELECT txn_id, amount, created_atFROM transactionsWHERE account_id = :accountAND DATE(created_at) = :dayORDER BY created_at;
Reads every row, then sorts what is left.
Response time
38.4 s
Before the fix
Execution plan
- Sortcreated_at
- Full table scantransactions
Sample schema and numbers.
Skills
- Languages
- Java
- Kotlin
- Python
- SQL
- TypeScript
- Frameworks
- Struts
- Apache Tomcat
- Django
- Flask
- JavaFX
- Data
- Informix
- PostgreSQL
- Neo4j
- Qdrant
- Execution plans
- Query tuning
- Engineering practice
- Profiling
- Code review
- SonarQube
- PMD
- SpotBugs
- Checkstyle
- PCI DSS
- AI tooling
- Claude Code
- Claude Agent SDK
- OpenAI SDK
- MCP servers
- Claude Code skills
- GraphRAG
- Platforms
- Linux
- Docker
- Podman
- Git
- GitHub Actions
- TeamCity
- Gradle
- Maven
Education
BS (Hons) Computer Science
Government College University (GCU), Lahore, Pakistan · 2019 – 2023
Coursework: Artificial Intelligence & Neural Networks, Machine Learning, Graph Theory, Parallel & Distributed Computing, Data Structures & Algorithms, Database Systems, Compiler Construction
Fellowship, Qasim Ali Shah Foundation
2022 – 2023 · One-year programme in communication and interpersonal skills.