Software Engineer — LLM Trainer & AI Agent Evaluator at Handshake AI | Previously Android Developer Intern at CodeAlpha
Java • Spring Boot • TypeScript • Distributed Systems
Building durable backends and AI agent evaluation platforms · LeetCode 2007+ (Top 3%)
LLM Trainer & AI Agent Evaluator at Handshake AI, designing multi-step software-engineering tasks, validation frameworks, and realistic scenarios to measure agent reasoning and coding ability.
Previously Android Developer Intern at CodeAlpha, shipping production-facing Kotlin apps with measurable UX and performance improvements.
- Backend engineering: Java, Spring Boot, REST APIs, JWT auth, idempotency, background workers
- Distributed systems: leases, fencing, crash recovery, workflow execution, job queues
- AI evaluation: benchmark authoring, hidden tests, agent scoring, comparison statistics
- Full-stack / product engineering: TypeScript, React, Next.js
- Mobile: Android (Java / Kotlin)
Core stack: Java, Spring Boot, TypeScript, React, PostgreSQL, Redis, Docker, Python
- Benchwright — Platform for building software-engineering benchmarks and evaluating coding agents. Publish versioned tasks with hidden tests, run patches in isolated sandboxes, and compare configurations with confidence intervals instead of a single pass rate. Live demo
- Stepwise — Durable workflow orchestrator in Java. DAG workflows on a worker pool with PostgreSQL as the source of truth — row-lock claims, heartbeated leases, fencing tokens, retries, and recovery so workers can crash without losing or double-applying work.
- LeetCode 2007+, Top 3% — hydra_19
- GeeksforGeeks 5-star, Top 0.8% — irshad
- Codeforces Expert, max rating 1636 — irshad
- CodeChef 4-star, max rating 1903+ — irshad
- LinkedIn: https://www.linkedin.com/in/mohd-irshad-1747b1238/
- GitHub: https://github.com/siddiqueirshad
- Portfolio: https://irshad-portfolio-seven.vercel.app
- Email: mohdirshad1306@gmail.com