👋

WORK

2024

Built and deployed AI voice agents from scratch using LLMs within a proprietary backend framework, integrated with LiveKit to power real-time conversational voice calls in production
Owned the end-to-end voice pipeline, focusing on post-call transcription, LLM-based conversation analysis, and structured insight generation
Implemented event-driven analytics workflows by provisioning and integrating AWS SQS and EC2, enabling scalable processing of transcripts and call outcomes
Designed LLM orchestration logic to analyze conversations, extract intent and metadata, and support continuous agent performance improvements

2024

Click to preview

Developed and deployed dynamic website for a luxury fashion house using Next.js, HTML, CSS, and JavaScript, focusing on enhancing user experience and visual appeal.
Managed image hosting for the site, ensuring seamless performance and fast load times for high-quality visuals.
Handled domain setup and implementation, ensuring the website's accessibility and smooth deployment.
Collaborated with the team to deliver a polished, user-friendly website, ensuring it aligned with the brand's premium aesthetic and functionality.

2024

Click to preview

Designed and implemented the company's landing site using React.js, HTML, CSS, and JavaScript, delivering an engaging user experience.
Contributed to the development of an AI-driven platform for generating instantaneous single-page websites based on user prompts.
Collaborated with the team to conceptualize and develop the website, ensuring seamless integration of AI-driven designs with smooth user interactions.
Integrated GSAP animations to elevate interactivity and visual appeal.
Created responsive website templates to enhance the platform's prototype.
Utilized Figma for design mockups, ensuring efficient transitions from concept to implementation.
Strengthened expertise in front-end development and team collaboration in a fast-paced startup environment.

PYTHON / DEEP LEARNING / NEURAL NETWORKS

Built a scalar-valued autograd engine implementing automatic differentiation from first principles
Implemented Value class with full operator overloading (+, *, -, /, **) and gradient tracking
Developed backward() method using topological sort for correct gradient propagation order
Created neural network primitives: Neuron, Layer, and MLP classes with tanh activation
Implemented chain rule backpropagation for arbitrary computational graphs
Built expression graph visualization for debugging forward/backward passes
Trained a multi-layer perceptron on binary classification with gradient descent optimization
Demonstrated understanding of how PyTorch/TensorFlow autograd works under the hood

PYTHON / PYTORCH / LANGUAGE MODELING

Built bigram and trigram character-level language models to generate name-like sequences
Constructed 2D count matrices in PyTorch to capture character transition statistics from 32K names dataset
Implemented probability distributions from raw counts with proper normalization
Used torch.multinomial for sampling from learned distributions with reproducible seeding
Visualized bigram frequency matrices using matplotlib to analyze character patterns
Extended bigram model to trigram for improved context-aware predictions

REACT NATIVE / NODE.JS / MONGODB

Architected zero-cost CDN using Cloudflare proxy with Backblaze B2, supporting 200+ users
Built cross-platform mobile app with React Native/Expo and real-time lyrics sync
Designed scalable REST API with batch processing, reducing API calls by 90%

2025

6.63 / 7 GPA

Consistently achieved high academic performance across all subjects, with High Distinctions and Distinctions in core areas such as Web Development, Programming, Cloud Application Development, and Software Engineering.
Specializing in Web and Mobile Application Development, Network Management, and Data Analytics for Cyber Security.
Actively engaged in IT Capstone Projects, focusing on real-world software development, network optimization, and cyber security solutions.