Yawar Mahmood

AI Engineer · Data Scientist · Applied Mathematician

I build ML systems that run in production: real-time computer vision live in retail stores, LLM agents that automate document and business workflows, and a consumer app scaled to 20,000+ users. M.Sc. in Industrial Mathematics & Statistics from NTNU. Currently freelancing and open to new roles.

Download CV Email LinkedIn GitHub

Highlights

10 stores

Real-time self-checkout system live for NorgesGruppen on NVIDIA Jetson edge devices

20,000+

Active users on the APIIR bike-fitting app, rated 4.5★ on the App Store

$300K+

Funding secured as CPO of APIIR

Experience

Freelance AI Engineer

Jun 2026 – Present
Self-employed
  • Automating slow, manual business processes for a restaurant client, including bookkeeping, purchasing, and marketing workflows, using Python, LLMs, and API integrations

AI Engineer

Mar 2025 – Jun 2026
Astar Consulting · Oslo
  • Developed a real-time self-checkout system for NorgesGruppen, now live in 10 stores on NVIDIA Jetson edge devices, built on NVIDIA DeepStream with anti-theft detection and Kalman filter-based multi-object tracking, tuned for in-store conditions
  • Fine-tuned DINOv3 with LoRA on augmented training data to cover rare products and difficult lighting conditions, improving accuracy of embedding-based product classification against a FAISS vector database of 10,000+ products
  • Built an agentic command-line system for Rapp Bomek with LangGraph and LangChain, where users give instructions in natural language and the agent carries out multi-step tasks on high-security door project documentation
  • Designed a multi-objective optimization engine for a startup, combining interpretable regression tree models with NSGA-II Pareto optimization to balance regulatory limits, CO₂ and cost in concrete recipes

CPO, Full-Stack Developer & Lead Data Scientist

Jun 2022 – Mar 2025
APIIR · Trondheim
  • Launched APIIR's B2C bike-fitting app from zero to 20,000+ active users across iOS & Android, rated 4.5★ on the App Store
  • Designed proprietary fitting algorithms achieving centimeter-level precision, fusing pose estimation with biomechanical modeling
  • Secured $300K+ in funding and drove end-to-end product strategy including roadmap, user research, and go-to-market
  • Architected the full-stack platform (Flutter, Django, cloud infrastructure)

Vice President & Head of Web Committee

Mar 2021 – Mar 2023
Nabla – NTNU Student Association for Physics & Mathematics · Trondheim
  • Owned the technical stack (website, IT operations, and digital services) for a 500+ member student organization
  • Co-led the board, responsible for strategy and member communications
  • Started and ran an office kiosk for members, and organized the association's 17 May breakfast end to end

Education

M.Sc. in Physics & Mathematics

2020 – 2026
NTNU · Specialization: Industrial Mathematics & Statistics
  • Thesis: Bayesian spatio-temporal model (INLA) of bus ridership in Trondheim from 1.7M+ boarding records, with well-calibrated uncertainty (90% predictive intervals reached 92% coverage)
  • Award: Best 1st Year Team, Nordic Collegiate Programming Contest (NCPC), 2020

Skills

Languages
PythonSQLRC++DartJavaScript
ML / AI
PyTorchComputer VisionLLMsLangGraphAI AgentsRAGFine-tuning (LoRA)NVIDIA DeepStreamEdge Deployment
Engineering
DjangoFastAPIFlutterReactDockerAzureGCP
Data
PostgreSQLMongoDBFAISSpandasREST APIs
Spoken
Norwegian (native)English (fluent)Urdu (fluent)German (basic)