AAcareercareerspentspentmakingmakingmodelsmodelssurvivesurvivecontactcontactwithwithproductionproduction

Senior AI Engineer

Sagaris · Remote, Virginia, USA

06/2026 - PRESENT

Building the core of SAGARIS, an agentic multi-touch revenue OS that automates outbound sales across email, SMS, LinkedIn, and social channels.

• Architect and build core modules of SAGARIS, an agentic multi-touch revenue OS that automates outbound sales across email, SMS, LinkedIn, and social channels.
• Engineered the multi-touch campaign orchestration engine driving 15+ step cross-channel sequences with per-contact channel deferral and reply-driven re-routing.
• Built the AI hyper-personalization engine generating context-aware outreach per prospect.
• Developed an in-house domain and mailbox warming subsystem for large-scale sending.
• Delivered agentic LLM features for the intelligent dialer with intent scoring and the real-time AI sales coach, orchestrated via MCP-based tooling.

• Python
• LLMs
• Agentic AI
• MCP
• FastAPI
• Node.js
• Email Infrastructure
• Deliverability

Lead AI Engineer

Orivfy · Remote, Texas, USA · Contract

01/2026 - PRESENT

Leading AI-generated-content detection across text, image, and video modalities, with explainable AI and provenance at the product core.

• Developed and enhanced the AI detection tool at Orivfy, focusing on improving accuracy in text detection models.
• Implemented passage-level and sentence-level AI classification to refine the model's performance.
• Integrated explainable AI and provenance into product architecture, ensuring transparency and reliability.
• Engineered real-time multi-modal fingerprint analysis across text, image, and video generators.

• Python
• PyTorch
• Transformers
• FastAPI
• MLOps
• Explainable AI

Software Developer

Ministry of Federal Education & Professional Training · Remote, Islamabad, Pakistan · Contract

02/2026 - 03/2026

Contract build of a CV-shortlisting desktop application automating candidate screening for government hiring pipelines.

• Built a CV-shortlisting desktop application to automate candidate screening for government hiring pipelines.
• Implemented CV splitting and OCR to parse bulk, multi-candidate documents into structured, machine-readable profiles.
• Developed a rubric-based analysis and ranking engine that scores and shortlists candidates against role-specific criteria.
• Generated per-candidate explainability detailing why each applicant passed or failed, enabling transparent and auditable decisions.

• Python
• OCR
• Document Parsing
• Desktop Application
• Explainable AI

AI Engineer II

National Center of Artificial Intelligence (NCAI) · Islamabad, Pakistan

09/2025 - 06/2026

Designed and deployed production-grade clinical AI, leading MedScribe, a real-time medical transcription and report-automation platform integrated with hospital EHR and PACS systems.

• Architected and delivered MedScribe, a production-grade clinical AI platform for real-time speech-to-text transcription and automated generation of structured, clinician-ready medical reports.
• Designed and optimized multilingual Automatic Speech Recognition pipelines, handling accent variability, domain-specific medical terminology, and noisy clinical environments.
• Implemented post-processing NLP pipelines for entity extraction, clinical summarization, and structured report generation aligned with healthcare documentation standards.
• Integrated Electronic Health Records and Picture Archiving and Communication Systems via secure, standards-compliant APIs, ensuring interoperability with existing hospital information systems.
• Enforced HIPAA-aligned security and compliance controls, including data encryption, access management, and audit logging across the entire data lifecycle.
• Designed scalable backend architectures using containerized microservices to support high-throughput, low-latency AI inference workloads.
• Led container orchestration strategies to ensure fault tolerance, horizontal scalability, and zero-downtime deployments in production environments.
• Optimized system performance through resource profiling, concurrency tuning, and service-level optimizations, reducing inference latency and improving system responsiveness.
• Implemented robust monitoring and logging mechanisms to track model performance, service health, and operational metrics in real time.
• Collaborated with cross-functional teams to translate clinical requirements into production-ready AI solutions while maintaining regulatory compliance.
• Contributed to architectural decisions involving service decoupling, API design, and deployment automation to improve maintainability and scalability.
• Led iterative model evaluation cycles, incorporating feedback from medical professionals to improve usability, accuracy, and trustworthiness of AI outputs.

• Python
• ASR/Speech Recognition
• NLP
• FastAPI
• EHR/PACS
• Docker
• Kubernetes
• AWS

Associate AI Engineer

National Center of Artificial Intelligence (NCAI) · Islamabad, Pakistan

12/2024 - 08/2025

Built an AI-powered medical imaging platform with 2D/3D DICOM viewers, computer-vision pipelines, and multimodal clinical AI for diagnostic workflows.

• Led the development of an AI-driven medical imaging platform leveraging the MERN stack and Python-based microservices, enabling modular, scalable system design for clinical-grade applications.
• Designed and implemented interactive 2D and 3D DICOM viewers using Cornerstone, supporting advanced radiological workflows and real-time manipulation of large-scale medical imaging data.
• Engineered high-performance rendering pipelines optimized for efficient handling of large DICOM datasets, ensuring low-latency visualization and smooth user interactions under clinical workloads.
• Built and integrated computer vision pipelines for medical image preprocessing, feature extraction, and AI-assisted interpretation across multiple imaging modalities.
• Fine-tuned large language models on healthcare-specific datasets for automated reporting on image modalities, incorporating tensor parallelism and pipeline parallelism to efficiently scale training and inference across multi-GPU environments.
• Designed multimodal inference pipelines combining imaging features and structured clinical context to generate consistent, clinician-ready diagnostic summaries.
• Developed secure, scalable RESTful APIs to support medical image retrieval, processing, and visualization across distributed systems and microservices.
• Implemented role-based access control, encryption, and secure data handling mechanisms to protect sensitive medical data and maintain healthcare compliance standards.
• Authored detailed technical documentation covering system architecture, data flows, and deployment strategies to support long-term maintainability and team onboarding.

• Python
• Computer Vision
• DICOM
• Cornerstone
• VLMs
• FastAPI
• OpenCV
• PyTorch
• MERN

Machine Learning Engineer

DevHawks · Islamabad, Pakistan

06/2023 - 11/2024

Built LLM-driven legal-intelligence systems for Pakistani courts, plus AI-safety work on deepfake detection and voice-cloning prevention.

• Developed an LLM-driven legal intelligence platform tailored for Pakistani courts and legal practitioners, enabling case building, law research, and AI-assisted legal document generation.
• Architected and implemented Retrieval-Augmented Generation pipelines supporting case-based and court-specific retrieval, optimized for Pakistani legal workflows.
• Designed end-to-end legal document ingestion and preprocessing pipelines, handling judgments, orders, petitions, and statutes from Pakistani courts with unstructured and noisy text.
• Led legal data creation and curation workflows, structuring court judgments, metadata, and citations to build high-quality datasets aligned with Pakistani jurisprudence.
• Fine-tuned Mistral 7B on a domain-specific legal corpus, improving legal reasoning, citation accuracy, and contextual consistency.
• Built semantic search and vector indexing pipelines to enhance precedent discovery while reducing hallucinations in LLM-generated outputs.
• Designed and deployed model inference APIs and backend services using Python and Node.js, enabling low-latency legal search and real-time generative responses.
• Developed automated legal document drafting and summarization pipelines, generating briefs, case summaries, and research notes aligned with lawyer workflows.
• Contributed to MediaGuard, a deepfake prevention and detection system, developing AI-based media authenticity verification pipelines using a CNN-LSTM hybrid model to detect manipulated audio, video, and image content.
• Developed audio-cloning prevention techniques by injecting imperceptible Gaussian noise into original audio recordings, disrupting cloned outputs by altering frequency patterns and preventing high-fidelity reproduction by generative audio models.

• Python
• LangChain
• RAG
• Mistral 7B
• LLMs
• NLP
• Computer Vision
• CNN-LSTM
• FastAPI
• Node.js

Machine Learning Intern

DevHawks

03/2023 - 05/2023

First ML role, fine-tuning transformer models for NLP and building computer-vision pipelines.

• Fine-tuned pre-trained models like BERT for various NLP tasks, enhancing language understanding capabilities.
• Engaged in Computer Vision projects, focusing on image classification, localization, and object detection.
• Explored classical machine learning and modern deep learning algorithms, applying hyperparameter tuning to optimize model performance.

• Python
• BERT
• Transformers
• Computer Vision
• Scikit-Learn
• PyTorch