Orchestration
Staged multi-agent pipelines with typed handoffs — LangGraph, Pydantic schema contracts, explicit stage ordering over emergent agent chatter.
About
How I work.
Engineer at Ember AI. Multi-agent orchestration and document-heavy retrieval, mostly for clients where a confident wrong answer is expensive.
I'm a software engineer at Ember AI, building production AI systems for clients in legal tech and media. Most of my work is multi-agent orchestration and document-heavy retrieval — the kind of systems where a confident wrong answer costs someone real money.
I work AI-first. Cursor and Claude Code are my primary implementation tools, and I write detailed specs as my unit of work rather than typing code directly. What that shifted isn't my speed — it's where the leverage sits. Architecture and data modeling matter more now, not less.
Most of what I've built lives in private client repositories, which is why the one system I can show in full is open source, end to end — retrieval, evals, and the failing cases left in the report.
Focus areas
Retrieval
Hybrid search over document-heavy corpora — vector plus full-text merged with RRF, structure-aware chunking, evals that keep their failures visible.
Infrastructure
Async Python services, Celery/Redis job pipelines, multi-provider LLM abstraction with failover, and the cost engineering that makes it affordable.
contact
- GitHub
- github.com/zainasher42
- /in/zain-asher-3960321b2
- Location
- Pakistan · open to relocation
- Timezone
- UTC+5 · overlaps EU mornings, US early
- Résumé
- Download PDF