The Person Behind
the Code.
I am Muaaz. A second-year CS undergrad who builds and ships real AI systems.
I work at the intersection of machine learning, backend engineering, and agent-driven architectures. I care about systems that do something useful in the real world, not just demos.
Most of what I build focuses on turning messy signals into structured decisions. Multi-agent workflows. LLM pipelines. Retrieval systems. Everything designed to be fast, reliable, and grounded.
I like taking ideas from research and pushing them into production. That means handling edge cases, scaling, and making sure the system holds up when it actually matters.
Outside of building, I write about what I learn, compete in hackathons, and explore how AI systems can be made more trustworthy and explainable.
Experience
Real work, real impact.
AI Intern
Hive Pro Inc.
Building multiple production AI systems with LangGraph-based multi-agent orchestration and RAG pipelines (project details under NDA).
Engineered durable, event-driven backend workflows using Temporal and NATS, exposed through FastAPI services.
Instrumented LLM pipelines with Langfuse for end-to-end tracing, evaluation, and observability.
Shipped a Next.js/React application now deployed and adopted across multiple internal teams, managed delivery via GitLab CI/CD.
Freelance AI Engineer
Client Project
Built multilingual AI voice calling system automating payment follow-ups and improving collection efficiency by 80%.
Integrated Twilio, Vapi, WebSockets, and FastAPI for real-time low-latency conversations.
Developed LLM pipelines extracting payment intent, sentiment, disputes, and call outcomes.
Engineered async backend workflows with scheduling, retries, and rate-limited orchestration.
Research Intern
NMIMS MPSTME
Co-authored an IEEE-format research paper on Janus, an intrusion detection system. Designed the full data pipeline over 2.8M network flow records and validated three adversarial model variants on live simulated traffic (paper under review).
Built end-to-end data pipelines for Indian legal judgment research, transforming unstructured court documents into structured datasets for downstream ML workflows.
Technologies
Tools and frameworks I use to build production AI systems.
Languages
Python
C++
Java
JavaScript
RustAgentic AI
LangChain
LangGraph
HuggingFaceML & Deep Learning
PyTorch
TensorFlow
Keras
Scikit-Learn
NumPy
Pandas
MatplotlibDatabases & Vector Stores
MongoDB
PostgreSQL
Neo4j
Redis
PineconeBackend
FastAPI
Go
Socket.IODevOps & Cloud
Docker
Kubernetes
AWS S3
Vercel
Render
RailwayFrontend
React
Next.js
Tailwind CSSCertifications
Courses and specializations I've completed.