Hi, I am
Chirag Malik
Building practical AI with purpose
Master's Graduate in Artificial Intelligence and I’m a full-stack developer who builds both modern web applications and intelligent AI systems. I work across frontend, backend, and AI, specializing in Generative AI, Large Language Models (LLMs), RAG systems, and AI agents to create complete, production-ready products
01. About Me
Hello! I'm Chirag Malik, a software engineer driven by the challenge of building intelligent systems that are as reliable as they are smart.
My journey began with a strong foundation in Computer Science, leading me to a Master's in Artificial Intelligence. However, I believe powerful AI needs robust engineering to be truly useful. I specialize in applying software rigor to AI—building automated testing pipelines and evaluation frameworks that ensure agents perform accurate, complex tasks in production.
Beyond models, I am a builder at heart. I approach every project with a language-agnostic full-stack mindset. Whether it's architecting scalable backends, designing efficient databases, or crafting intuitive user interfaces, I focus on choosing the right tool for the job to bridge the gap between complex AI research and polished, production-ready software.
02. Where I've Worked
AI & Full Stack Developer @ Mxpert
Sep 2025 - Dec 2025
- Built an automated testing pipeline for conversational AI agents using Python, ML methods, and OpenAI APIs, validating accuracy, pricing logic, and multi turn response quality across real service scenarios.
- Designed a multi stage evaluation framework using cosine similarity, LLM scoring, and domain checklists to measure responses across 8+ criteria, helping improve our AI agent model to achieve over 98 percent accuracy.
- Implemented dynamic state slot memory systems to preserve key attributes like service, price, and duration, significantly improving multi-turn contextual consistency.
- Improved customer data workflows on the frontend by enhancing React and TypeScript components, making internal tools more intuitive and reducing steps needed to access services and information.
- Strengthened backend integrations and data retrieval by extending and optimizing API endpoints, ensuring smoother communication between frontend views and backend systems while enabling new product features.
AI Developer @ Genesis
May 2025 - Aug 2025
- Built a full stack RAG chatbot using Llama 3, FastAPI, and React, achieving highly accurate, context aware responses across a dynamic knowledge base of 150+ documents.
- Designed a scalable pipeline to parse PDF/DOCX files, perform chunking and embedding, and store vectors in Qdrant, enabling low latency semantic search (<500ms).
- Developed a modular backend with PostgreSQL, Firebase Auth, and REST APIs to handle user sessions, chat history, and document uploads.
- Created a responsive React and Tailwind frontend with real time chat UI, integrated with Firebase Auth and Context API for seamless state management.
- Validated system performance and accuracy through rigorous testing with a diverse queries confirming the model's ability to minimize hallucinations and deliver consistently relevant answers
Undergraduate AI Research Assistant @ MIET
September 2022 - February 2023
- Built a CNN-based brain tumor detection tool using MRI data, achieving 98.25% training and 97% validation accuracy
- Improved model generalization by applying data augmentation and cross-validation, reducing overfitting by 15%
- Published results at an IEEE conference and presented to 100+ attendees, showcasing advances in AI for medical diagnostics
03. Some Things I've Built
Featured Project
HeartHush - AI Mental Wellness Platform
A private mental wellness platform with an AI chatbot built using RAG on world-class psychological research and fine-tuned LLMs trained on therapist style responses. It offers safe, context-aware conversations, emotion and mood analysis, and evidence-based exercises, tools and interactive mental health games. Designed with strict user confidentiality and a seamless, responsive chat experience across the entire platform.
- RAG
- LLM Fine-Tuning
- Python
- Postgres + VectorDB
- React
- Next.js
Featured Project
Servira - Business Automation Platform
Servira helps service professionals streamline scheduling, client management, invoicing and payments through automated booking, a built-in CRM, AI-assisted workflows and secure Stripe billing. With an intuitive dashboard, calendar scheduling, real-time insights and self-service client pages, Servira provides a smooth end-to-end workflow that saves time, automates operations and supports business growth.
- React
- TypeScript
- Python
- OpenAI API
- Stripe
Featured Project
Fine-Tuning Pipeline for Local LLMs
Built a general-purpose fine-tuning pipeline for Qwen 3 (0.6B) and other local LLMs using LoRA (PEFT) and the Unsloth framework with 4-bit quantization (QLoRA-style), enabling efficient domain adaptation on custom datasets. Includes Hugging Face Datasets integration for instruction-style preprocessing and exports models in GGUF format for fast, offline inference via Ollama.
- Python
- PyTorch
- LoRA
- PEFT
- QLoRA
- Unsloth
04. What's Next?
Get In Touch
I’m currently looking for new opportunities, my inbox is always open. Whether you have a role in mind, a project to collaborate on, or just want to connect - feel free to reach out.