Select an interaction mode to engage with my AI assistant. Real-time voice or text-based communication protocols available.
Hi, I'm Ankit, a Computer Engineering Student specializing in high-performance production ready AI architectures and scalable web systems that actually brings value to real-world applications.
I'm a backend developer and AI/ML engineer building production-ready systems that scale. Shipping high-performing Discord bot scaling over 1lakh+ users and 2k+ servers, from designing business-ready automations to engineering real-time across computer vision and GenAI involving deep learning inference, embeddings, clustering, and retrieval-augmented workflows. I love turning noisy real-world signals into dependable products by optimizing for performance, and ease of use.
My academic journey in computer science and technology.
Bachelor of Technology in Computer Science
Higher Secondary Education
Secondary Education
Building production systems and shipping real-world solutions across AI and web.
Applied AI/ML Engineer
Implemented embedding-driven semantic indexing with clustering (KMeans) to surface latent themes from high-volume text signals, then layered retrieval-augmented generation (RAG) for grounded summarization and query-time synthesis—cutting insight turnaround time by 37% and improving cluster/label quality by 21% in iterative evaluation. Built a fully local, streaming speech stack (wake-word → incremental ASR → low-latency LLM inference → neural TTS), adding caching/buffering, lightweight batching, and profiling-led optimizations to reduce end-to-end latency by 45% and improve response consistency under continuous real-time interaction.
Web Developer & AI Automation Engineer
Shipped multiple client-facing web experiences as a freelancer to various clients and businesses, designing sleek, performant, and scalable frontends, modern build tooling, and a polished UI/UX layer backed by a reusable component architecture. Also engineered conversational automation systems that interpret appointment intents from chat with a massive growth in business almost by 37% and a significant reduction in manual scheduling workload.
Discord Bot Developer
Built and scaled production-grade Discord bots with a focus on reliability, performance, and monetization. Most notably my best performing bot - Sync Music, a verified feature-rich Discord music bot built in TypeScript on discord.js v14, backed by MongoDB, and powered by Lavalink via erela.js for high-quality audio streaming. Successfully grew Sync Music to over 1.5k+ servers, turning it into a source of passive income through premium/vote-driven features and distribution.
Areas where I focus my expertise and deliver measurable impact.
Analyze and interpret visual data with advanced convolutional neural networks and transformer architectures.
End-to-end development for scalable, modern applications using distributed systems and real-time data flows.
Streamline complex enterprise processes with intelligent, autonomous systems and LLM orchestration.
A selection of projects spanning web systems, AI pipelines, and scalable infrastructure.
Verified Discord music bot used by 200k+ users across 2k+ servers. Built a scalable Discord.js v14 + TypeScript architecture with a pluggable command/event layer, MongoDB-backed persistence for user-music data, and Lavalink-powered audio streaming paired with shokaku wrappers for interactive controls, while supporting monetization via vote-locked and premium feature gates.
Implemented an end-to-end comment ingestion + preprocessing pipeline, embedding-based semantic indexing, and KMeans theme discovery, then added a RAG-style chat layer to convert unstructured feedback into concise, queryable insights, improving insight discovery speed by ~30% and reducing manual analysis effort by ~50% in iterative runs.
Developed a real-time computer vision attention monitoring system in Python using MediaPipe and OpenCV, implementing 468-point facial landmark tracking and Eye Aspect Ratio (EAR)-based blink/attention scoring to reach 94.3% detection accuracy.
Seeking opportunities at the intersection of infrastructure and artificial intelligence.
In the labyrinth of neural networks, data isn't just stored; it's synthesized into a semantic map of human knowledge...
Beyond the hype of "digital brains" lies a brutal process of gradient descent and weight optimization that mirrors...
When you ask AI a question, millions of matrix multiplications cascade through layers. Here's what inference actually looks like...
In 2017, a research paper quietly changed the world. It replaced recurrence with attention mechanism, paving the way...
Tokens, embeddings, and vector spaces: the hidden translation layer that converts human language into mathematical representations...
Emergent properties in large scale models defy simple logic. It's not just about more data, but about phase transitions...
Pre-training gives AI knowledge, but fine-tuning gives it purpose. The secret sauce that turns a language model into an assistant...
RLHF and human feedback loops shape AI behavior. The invisible army of annotators teaching machines to be helpful and harmless...
