Software & Machine Learning Engineer
I design and implement robust, high-performance software systems. As a Computer Science graduate from FAST-NUCES, my work bridges the gap between complex algorithms and intuitive, functional user experiences as a Software & Machine Learning Engineer.
Whether it is building mathematical optimization engines, implementing RAG pipelines with vector databases, or training face recognition models for automated media workflows, I focus on system efficiency and correctness.
A RAG-based resume advisor and career optimizer. Utilizes a FAISS vector database with sentence-transformers embeddings to match user CVs against live job postings. Computes skill overlaps, calculates a custom resume score (0-100), and interfaces with Google Gemini to generate LaTeX-formatted resume rewrite suggestions and personalized cover letters.
A multi-task support triage engine and explainability dashboard modeled for Riot Games player support. Features a Platt-calibrated LinearSVC classifier reducing mean Brier score probability error by 99.85%, an out-of-distribution (OOD) query guardrail, a GPU-accelerated XGBoost priority ranking model, and real-time linear hyperplane feature attribution (token contributions) served via a high-throughput FastAPI REST backend.
An end-to-end computer vision and NLP system for automated document classification, spatial text extraction, and structured field parsing. Powered by a PyTorch ResNet18 transfer learning classifier, spatial EasyOCR bounding-box detection, category-specific regex key-value extraction, an interactive canvas visualizer with real-time field highlights, and persistent SQLite audit logging.
A full-stack Wordle-style guessing game for League of Legends. Architected with Next.js and TypeScript featuring custom SVG sprites and canvas elements, integrated with a serverless Supabase PostgreSQL database with RLS policies and custom SQL RPCs to aggregate daily solver statistics, a build-time pre-fetch data pipeline in Node.js utilizing Riot Games' API, and optimized static asset deployment on Vercel.
A high-performance layout-preserving PDF to Word DOCX reconstruction system. Features real-time Server-Sent Events (SSE) telemetry tracking a 4-stage conversion pipeline (upload, layout topology, page-by-page reconstruction, and packaging), client-side preflight page range extraction with PDF.js, and an automated zero-retention ephemeral cleanup daemon.
An AI-powered full-stack photography booking platform and marketplace. Features a mathematical photographer-matching optimization engine (ILP via PuLP), a secure Stripe escrow payment system, real-time WebSocket chat, and an AI smart album builder utilizing InsightFace ArcFace embeddings, DBSCAN clustering, CLIP zero-shot mood detection, and OpenCV image quality analysis.
An AI-powered competitive analysis and scouting dashboard built for Valorant esports coaching staff. Incorporates a natural language chat interface leveraging Google Gemini to dynamically generate and execute complex SQL queries over a DuckDB database of match history, delivering live team overview, player stats, map analysis, and exploitable weakness reports.
I am a Computer Science graduate from FAST-NUCES and Machine Learning Engineer with a passion for building complex, data-driven systems, computer vision pipelines, and end-to-end full-stack applications. My approach centers on combining robust backend engineering with responsive, intuitive frontend interfaces to deliver seamless, production-ready solutions.
My hands-on experience spans the entire application stack: from designing scalable database schemas, developing fast REST/WebSocket APIs, and setting up vector databases for RAG pipelines, to building dynamic user dashboards and fine-tuning computer vision and NLP models.
I thrive on building high-performance software that solves concrete, real-world problems—whether it is building AI-driven scouting dashboards for esports teams, RAG advisors calibrated against live job feeds, Platt-calibrated support triage classifiers, or spatial OCR document extraction engines.
Currently open to new opportunities. Whether you have a question, a project proposal, or just want to say hi, my inbox is always open.