About
Full Stack Engineer with 2+ years of experience building and deploying production-scale applications used by 50–200 users. Skilled in Angular/React, Spring Boot, and Flask for developing scalable REST APIs and real-time systems using WebSockets. Experienced in Docker-based deployments on GCP Cloud Run, microservices architecture, and CI/CD workflows. Strong background in integrating AI systems (LLM/RAG, computer vision) into full-stack applications.
Skills & Expertise (35)
Work Experience
Machine Learning Engineer
Manomaya AI Systems
Oct 2024 - Mar 2025
Integrated custom YOLOv8 ML pipelines for aircraft, vehicle, and passenger detection into production systems, improving detection accuracy from 80% → 96% mAP. Designed turnaround event logging for airports and multi-camera passenger counting with ByteTrack, fully deployed in live environments. Architected an end-to-end passenger counting system using ByteTrack + YOLOv8 with line-crossing detection across 4 simultaneous camera streams, deployed against live RTSP airport cameras in production. Optimized inference via TensorRT (.pt to .engine conversion) on remote GPU hardware, reducing latency from 15ms to 3ms, a 5× speedup; researched and ruled out VLM alternatives (LLaVA, Qwen-VL, BLIP-2) after documenting failure modes. Built React frontends for two production airport AI systems, enabling video upload, real-time inference triggering, results visualization, and per-camera passenger count display for non-technical airport operations teams. Developed Flask APIs orchestrating ML inference (YOLO/ByteTrack) with structured outputs. Architected 3-container Docker microservices systems (ML inference + Flask backend + React UI) for both products, enabling clean separation of concerns and portable deployment across environments.
Full Stack Engineer
Manomaya AI Systems
Apr 2024 - Present
Led development of a recruitment platform supporting ~50 active users, with REST APIs for candidate management, job postings, and search filtering. Built a RAG-powered recruiter AI helpdesk chatbot using Flask and GPT-4o-mini, implementing intent classification with confidence scoring, knowledge base retrieval fallback, context-bound prompt construction, and strict JSON schema validation on all LLM responses. Built Angular chatbot UI with conversation state management, query handling, loading states, and fallback error handling. Implemented best-effort transcript persistence for recruiter and assistant messages, designed as a non-blocking write so transcript failures never impact the live user response path. Managed development using Git feature branches and pull requests, validated all REST APIs through Postman integration testing and performed end-to-end manual testing across recruiter workflows before release. Shipped DeWiser end-to-end, a production system actively monitoring 50+ devices for 50–100 users, featuring real-time CPU, memory, and hardware dashboards, anomaly detection alerts, and a structured event logging system. Engineered the DeWiser Windows Service as a persistent background process on production machines, handling system event tracking, automated log rotation, hardware change detection, and offline queue sync to ensure zero data loss during connectivity failures. Built and deployed scalable Spring Boot REST APIs (JPA/Hibernate) via Docker + GCP Cloud Run. Designed DeWiser Web in Angular with WebSocket integration enabling real-time remote monitoring of 50+ production devices, featuring live dashboards, system metrics visualization, remote session management, and a ticketing system from a single interface. Ensured production stability through Git PR workflows and end-to-end API integration testing.
Education
B.E. in Electronics and Communication Engineering - GM Institute of Technology, Davangere
2019 - 2023 · Afghanistan
