Available for new engagements
Multi-agent workflows. 8+ years full-stack. Currently shipping LLM features at Jobandtalent.
About
I'm a software engineer building AI-native products. Currently at Jobandtalent in Spain, I ship LLM features for hiring and payroll — vacancy generators, screening agents, and the observability that keeps them honest in production.
Before that I led Sendoso's Fulfilment team through scale-up: rearchitecting the order monolith around Kafka, redesigning Salesforce integrations to live within rate limits, and steering an engineering team that owned warehouse fulfilment across the platform.
I work async-first from Pakistan on engagements anywhere in the world. If you're shipping AI features and want a senior IC who's done it before — across the stack, with the operational instincts of a former lead — let's talk.
Services
Four ways to work together. Each one is built to drop into a team that's already moving — not slow you down with discovery rounds.
Production LLM features end-to-end. Vertex AI, OpenAI/Anthropic APIs, dynamic model routing, multilingual generation, full prompt/cost observability with Langfuse-style instrumentation.
Orchestrator + specialized sub-agents (Staff Engineer, System Architect, Security Auditor, QA, Devil's Advocate) running in parallel Git worktrees with human approval gates. For teams that want to ship faster without losing review quality.
Next.js / TypeScript / React on the front, Kotlin / Ruby on Rails / PostgreSQL on the back. AWS, Trigger.dev, Drizzle ORM. Greenfield builds and rescues.
Embed with your team to design AI workflows, stand up Claude Code / MCP infrastructure, or untangle a thorny system.
Selected Work
Four engagements where the work moved real numbers in production.
Shipped an LLM-powered job description generator on Vertex AI with dynamic model routing, native multilingual output, and end-to-end Langfuse observability over prompts, latency, and cost.
Integrated Clara (an AI screening agent) into recruiter views and consolidated candidate data across upstream sources into a single funnel.
Led the team owning order creation, orchestration, and fulfilment across warehouses; designed Kafka event-driven decoupling of the order monolith.
Redesigned long-running jobs to respect API rate limits, eliminating daily integration failures across the Salesforce sync surface.
Approach
An orchestrator splits the work, specialized sub-agents run in parallel Git worktrees, and a human reviews between phases. You get speed without losing the review quality you'd expect from a senior team.
A planning agent scopes the task, breaks it into independent units, and decides which specialized sub-agents to dispatch. Output is a written plan — not a black box.
Staff Engineer, System Architect, Security Auditor, QA, and Devil's Advocate each run in their own Git worktree. Independent work happens concurrently with zero merge contention.
You review at deliberate checkpoints — plan, draft, final — instead of every keystroke. Your eyes stay on the decisions that matter; the agents handle the mechanical parts.
Worktrees converge into a single review-ready branch with passing tests, threat-model notes, and a PR description that explains the why — not just the diff.
Tech
Tools I use day-to-day. New ones welcome — most of these I picked up because a project needed them.
What people say
His willingness to collaborate and support his colleagues fosters a very positive working environment. Having Touseef in the team is a real pleasure!
Touseef is a person who brings great energy to a team. He approaches issues not with a “can this be solved” attitude but “how quickly can this be solved” attitude.
Touseef is very driven and a very hard working engineer, often working around the clock to build new innovative features for Sendoso or addressing critical bugs.
Contact
Tell me a little about what you're working on. I read every message and reply within two business days.
