Production AI systems.
Vision, LLMs & infrastructure.
Built to ship.

AI Engineer

Symufolk · Oct 2025 — Present · Lahore, Pakistan

• Designed distributed inference architecture with CPU-based backend and GPU-hosted model services (Docker), orchestrated via Redis queues and background workers; enabled on-demand GPU utilization, reducing infrastructure costs by ~60–80% and improving system scalability.
• Leveraged vision models (SAM/YOLO) to auto-label video frames, accelerating annotation throughput and eliminating the need to hire 5 additional annotators, saving ~$2K/month (~$24K/year) while scaling dataset generation.
• Extended open-source Label Studio to build an enterprise-grade annotation platform on GCP; implemented RBAC, review workflows, and custom storage integration, enabling 100+ daily annotators across 50TB+ data while eliminating ~$600/month (~$7K/year) in licensing costs.
• Designed and executed a benchmarking framework for vision models (Grounded SAM, SAM3, YOLOv8/v10, ByteTrack, AprilTag), evaluating accuracy, latency (10–30Hz), and reliability, enabling safe model selection for real-time deployment.
• Built a scalable scraping and pricing pipeline (10K+ products, ~4-hour runtime), enabling automated, rule-based price updates and real-time market competitiveness.

Data Annotator

Motive · Aug 2025 — Oct 2025 · Remote, Pakistan

• Annotated 1,000+ image and video samples daily to support supervised learning pipelines for autonomous driving perception models.
• Achieved 97%+ annotation accuracy by adhering to strict QA standards, labeling guidelines, and multi-stage review processes.
• Conducted dataset validation and consistency checks to reduce labeling noise and improve training data reliability.
• Applied structured annotation schemas within a large-scale ML data pipeline environment while meeting strict delivery SLAs.

Software Engineer

Aspire Analytica · Jul 2023 — Jun 2025 · Lahore, Pakistan

• Engineered an agentic AI financial reconciliation platform (RAG, tool-augmented agents, API orchestration), reducing a 2-week manual workflow to 24–36 hours (~85% faster) while improving audit consistency and reporting accuracy.
• Led development of LLM-powered conversational and voice agents (OpenAI, LangChain, STT/TTS) across restaurant, automotive, and real estate domains, reducing support workload by 60–70% and operational costs by 40–50%.
• Built browser-based RPA and Google Apps Script automation pipelines, eliminating 24/7 manual scheduling dependencies and reducing repetitive workload by 60–90%.
• Architected backend systems (Python, FastAPI) for a multi-entity ERP platform (POS, inventory, HR, accounting), enabling inter-company reconciliation across 5+ business units, reducing accounting overhead by ~35%, and supporting 1K+ daily transactions.

Inbound Customer Service Representative

Tribe Consulting (Pvt.) Ltd. · Apr 2022 — May 2023 · Lahore, Punjab, Pakistan

Provided customer support in a high-volume inbound call environment, managing order processing, issue resolution, and operational support while consistently meeting service quality and performance targets.

• Handled 80–100 inbound customer calls daily, assisting with order placement, account inquiries, and service requests in a fast-paced operational setting.
• Investigated and resolved customer issues, escalating complex cases to specialized teams to ensure timely and accurate resolution.
• Maintained a 95%+ quality assurance score while working under strict SLA and productivity benchmarks.
• Documented customer interactions and followed standardized workflows to ensure operational consistency and compliance.
• Contributed to positive customer experience outcomes through effective communication, problem-solving, and real-time decision-making.