Python-first Full-Stack & AI Engineer
Leading AI & backend engineering since 2020 — production AI agents, multi-tenant SaaS, and intelligent automation.
Replaced a legacy ASP.NET (c. 2005) retail-merchandising platform serving 10,000+ stores, 250+ field associates, and clients like Target & Kroger.
→ Rebuilt in 8 months as a Django SaaS on SaaS Pegasus. 10M+ records migrated with zero data loss, 26 Django apps, 15k+ daily API requests under 200ms, DB perf improved 45%.
Full recruitment automation: resume parsing, JD-based scoring, and a live AI-presenter interview bot with anti-cheat.
→ Sole engineer. Built end-to-end pipeline: BarkTTS voice cloning → Wav2Lip/MuseTalk lip sync → realtime interview bot with gaze tracking and tab-lock anti-cheat. Deployed across 4 enterprise clients screening 500+ candidates/month.
Production-grade wound segmentation, stage classification, and clinical extraction — running on-device on mobile.
→ Custom UNet + EfficientNet-B4 with hybrid Dice/Focal loss reaching 95%+ accuracy. INT8-quantized for on-device inference (4× smaller, 3× faster). FastAPI inference backend.
Saudi hospital needed an autonomous voice agent for booking, cancellation, rescheduling, and doctor availability — not an IVR.
→ Fully agentic LLM with tool-calling: booking, cancellation, availability, FAQ. Twilio + OpenAI Realtime API for natural multi-turn voice. Handling 300+ calls/day with sub-2s response latency, replacing 2 FTE call-center seats.
Let any business spin up an embeddable AI chatbot grounded on their own documents, with human-agent escalation.
→ RBAC multi-tenancy, vector-DB RAG over uploaded docs, iframe embed, tool-calling escalation to live human, voice chat. Shipped to 6 paying tenants indexing 12,000+ documents in production.
Legacy WordPress took ~10s to load 1,000 listings against 10M+ housing records. Unusable.
→ Rebuilt as Django + PostgreSQL with Pandas bulk ingest and Redis cache on AWS EC2. 10s → 200–300ms (~50× faster). Employee of the Month.
An interactive map of how my employers, projects, and skills relate. Hover any node to highlight its connections, or click a role or project to jump to its section.
I want to recommend Mohsan for his excellent skills in development and research. He was one of the best students in the Software Engineering Department at the University of Lahore. Mohsan showed a strong understanding of the subject and always solved problems well. He works great with others and is always eager to learn more. He would be an excellent fit for any opportunities in this field.
Mohsan consistently demonstrated a firm grasp of NLP fundamentals and applied them to practical scenarios with impressive dedication. For their final-year project on person re-identification, they showcased incredible expertise in machine learning — designing a robust system that accurately identifies individuals across multiple cameras. What stood out was their proactive problem-solving: they analysed research papers, adapted algorithms, and integrated components into a cohesive system. An excellent team player with strong communication and leadership qualities. I wholeheartedly recommend Mohsan and have no doubt they will make significant contributions to NLP and machine learning.
Mohsan Yaseen has tremendous problem-solving skills. We studied together for four years and worked on several joint-venture projects. His Python programming skills are very good, as well as his database skills. I surely recommend him for any kind of AI-based model training as well as for the development of complex backend projects.
Person Re-Identification web app — Transformer-based Re-ID fine-tuned on a custom dataset; React + FastAPI; multi-camera tracking with live & DVR/NVR feeds.
Active MSc-level research in medical computer vision — wound segmentation with UNet + EfficientNet-B4 and Re-ID transformer architectures for multi-camera tracking.
Open to remote full-time and contract roles in AI/backend engineering — and to Riyadh, Saudi Arabia based senior/lead positions. Quick to respond.